Top stories

A reference to "frantic-ferret," a model whose developer was not disclosed, appeared in the model catalog of the AI coding tool OpenCode on Nov. 8 (Nov. 9 Japan time). An X account that tracks new models detected it, but the post itself said availability and specifications were unconfirmed. OpenCode's official usage statistics recorded about 9.9 billion tokens and 347 users on Nov. 8 alone. As of Nov. 12, the model was not listed in the official catalog or documentation, and the statistics still showed its developer as "Unknown." No company has announced what the model is.

Google said on Nov. 11 (Nov. 12 Japan time) on its official X account that developers can now connect sandboxes (isolated execution environments) they set up themselves to Managed Agents, the AI agent feature of the Gemini API. Tools that had run in Linux sandboxes provided by Google can now run on a company's own computing infrastructure. The connection uses outbound HTTPS from the open-source Gemini API CLI, so no inbound firewall ports are needed. The feature is in preview, and Google says its specifications may change. The official documentation does not mention pricing for computing resources in self-hosted environments.

OpenAI said on Nov. 11 (Nov. 12 Japan time) through ChatGPT's official account on X that the personal finance feature "Finances" is now available in the ChatGPT desktop app. Users get it by updating the app, and it is available to Free, Go, Plus and Pro users in the U.S. The post did not say which operating systems are supported. As of the morning of Nov. 12 Japan time, OpenAI's help page still described Finances as available on web, iOS and Android in the U.S., with no mention of the desktop app.

Google is moving to add "xhigh" to the setting that controls reasoning depth (thinking level) in its developer-facing Gemini API, according to changes in its official SDKs (software development kits). The values are ordered minimal, low, medium, high and xhigh, and xhigh appears to be a level above "high," the deepest so far. The addition also appeared in release notes for the official Python, JavaScript and Java SDKs dated Nov. 10. Google has not announced it on its official blog or official X accounts, so the information remains an unannounced leak. Supported models, pricing and whether the value can actually be called have not been disclosed.

Google said on Nov. 9 that developers can now build apps using its mapping service "Google Maps Platform" inside "Google AI Studio," its service for developers who build apps with AI. When a user describes the app they want in text, AI Studio handles the map setup and prepares the API key, and an interactive map appears in the preview. The prototyping "Maps Demo Key" can be used without a credit card but is limited to 100 calls per API per day, and some APIs are excluded. Production use or more calls require linking a Google Cloud billing account, with standard charges applying. The announcement did not state pricing amounts.

Voice AI company ElevenLabs on Nov. 9 announced "ElevenAgents Campaign Manager," a new feature of its voice agent service "ElevenAgents" for "outbound campaigns" in which companies call customers to drive sales, contract renewals, collections and other work. Given contacts and a goal such as qualifying a lead or confirming an appointment, the agent places calls, calls back later if it does not get through, and carries the content of earlier conversations into the next contact. Calling hours, maximum attempts and do-not-contact lists can be set, and progress is recorded for each contact. Existing customers can use it from the same day. It supports voice and messaging on phone and WhatsApp, and the company said email and SMS will be supported in the future. The official blog does not state pricing.

Google said on Nov. 9 it has added new systems, "TPU 8i Max" and "TPU 8tx," to the 8th generation of "TPU," its in-house semiconductors for AI. TPU 8i Max puts two TPU 8i chips in one package, doubling compute, SRAM and HBM bandwidth compared with TPU 8i. TPU 8tx pairs the same silicon as TPU 8t with AMD CPUs and is configured for heavily regulated companies and governments, protecting data and model weights in use with hardware. Together with the two products from April, the company said it is the first time it has released four TPU systems in a single year. Google did not disclose pricing, availability dates or regions.

Anthropic released the AI models "Claude Fable 5.2" and "Claude Mythos 5.2," designed for the most difficult technical challenges, in an official blog post dated Nov. 9. The two share the same underlying model; Fable, which has strong safeguards for cybersecurity and biology, is available to users on the Pro, Max, Ultra, Team and Enterprise plans and through the API. Mythos is for organizations that pass a review. API pricing is $8 per million input tokens and $40 per million output tokens, down from $10 input and $50 output for 5.1. Anthropic said the model scored 49.1% on FrontierCode 2.0, an evaluation of development work (with reasoning at Xhigh), above Claude Opus 5.5's 45.5%. In an evaluation by third party Artificial Analysis (AA), it ranked first on the overall index with 65, but its cost per task at the deepest reasoning setting was $26.5, about 2.6 times Opus 5.5's $10.0.

OpenAI on Nov. 9 made it possible to connect multiple local computers to its always-on AI agent "dots." Users can choose which computer to use for each task. The company gave the example of having a dedicated machine build and test an app while the user keeps coding on a laptop. According to the official documentation, one thread uses one computer at a time, and separate tasks run in separate threads for each computer.

Meta released "Muse Ember," its smallest and most efficient AI model, in an official blog post dated Nov. 9. It is aimed at the tool calls, fine-grained coding tasks and long-document reading that AI agents repeat all day. It became available the same day in the coding tool "Muse Code" and the "Meta Model API," with API pricing of $0.10 per million input tokens and $0.50 per million output tokens. The context window is 1 million tokens, and output can be up to 512,000 tokens (512K). Citing Artificial Analysis results, Meta said it delivers more performance per dollar than Anthropic's "Claude Haiku 5.5" and OpenAI's "GPT-6 Luna" at every reasoning setting. However, Haiku 5.5 scored higher on FrontierCode 2.0-AA, which grades real-world development work. CEO Mark Zuckerberg posted on X on Nov. 9 (Nov. 10 Japan time) that Muse Ember and other models will be offered on Google Cloud.

Video generation AI company Runway announced on its official X account on Nov. 9 (Nov. 10 Japan time) a new feature, "Sightline," that creates a 360-degree panorama of any location and lets users frame shots from any angle within it. The surrounding environment stays the same at whatever angle a shot is taken, and the panorama can be exported for 3D to use as a scene's environment. The company said users can get started from "App Studio," its system for building production apps released in October. It did not give pricing or eligible plans.

SpaceXSI (SpaceX Super Intelligence) has released version "1.0" of its AI agent "Grok Bot." Its official news post calls it "a major update to our AI teammates." The update adds "Team Bots," which let a team share the same bot; collaboration in Slack and Microsoft Teams; joining meetings on Google Meet, Zoom and Teams; proactive offers to take on work; and learning a team's preferences and decisions. Team Bots are available on the Team and Enterprise plans, and meetings on the Team, Enterprise and Ultra plans.

Voice AI company ElevenLabs unveiled "ElevenAI," an AI assistant that helps employees with their work, at the "ElevenLabs Summit" it held in New York on Nov. 11, and opened free early access the same day. It lets users dictate into any app, takes meeting notes and provides live assistance during calls. For "ElevenAgents," its customer-facing voice agents, the company said it will add "Memory," which keeps records for each customer, and "Routines," which automatically handles tasks before and after conversations. It showed "Interaction Engine," which manages the flow of conversations, as a preview. It also launched a partner program and introduced a "Marketplace" listing agents built by partners. Purchases on the Marketplace will be possible from next year. The same day, it announced an expanded partnership with NVIDIA and made NVIDIA's model "Nemotron 3.5 Super VL" available in ElevenAgents. ElevenLabs' own evaluation found its cost 70% lower than other major open models.

Elon Musk, who leads SpaceXSI (SpaceX Super Intelligence), posted on X on Nov. 10 that the next model, "Grok 4.8," has finished pretraining and entered the reinforcement learning stage. He said it will begin rolling out gradually to users of the AI agent "Grok Bot" later this week to see how it does. Asked how it compares with Anthropic's Claude Opus 5.5, he replied that it is "not as good." As of the morning of Nov. 12, Grok 4.8 was not listed on the company's official website or in its developer documentation, and the latest official model remains Grok 4.7.

Google DeepMind announced "AlphaProof Code," a new framework for formal verification, on Nov. 10 (Nov. 10 Japan time). Under human guidance, AI agents mathematically proved key security properties, including forward secrecy and authentication, of "libsignal," the reference implementation of the cryptographic protocol of the messaging app Signal, at the level of compiled machine code. According to the paper, the proofs generated by the agents run to about 6.5 million lines, and the work took a team of five four weeks. The target is an executable run from a CLI, and low-level cryptographic operations are treated as assumptions.

Anthropic said on Nov. 10 it has begun its roadshow presenting to institutional investors ahead of a Nasdaq listing. It is offering 825 million Class A shares at an expected price of $70 to $78 per share. According to an amended registration statement filed the same day, seven companies including Google and NVIDIA will buy a total of $26 billion of non-voting shares at the offering price immediately after the listing. The company's net proceeds are expected to be about $84.6 billion at $74, the midpoint of the price range. The registration statement has not yet become effective, and no listing date has been given.

Voice AI company ElevenLabs announced new speech-to-text models "Scribe v3" (for recorded audio) and "Scribe v3 Live" (for real time) on its official blog on Nov. 10 and made them available through its API. The company said the recorded-audio model makes 27% fewer word errors than the previous Scribe v2 and 12% fewer speaker-separation errors. It said Live cuts semantic errors by 43% compared with the previous version. All figures are the company's own measurements. API pricing per hour is $0.20 for the recorded-audio model and $0.22 for Live, listed in the same rows as the previous versions in the price table. Live is a speech recognition model and a separate product from "Eleven Realtime," the company's model for voice agents.

Google has released "Gemini Decision" (gemini-decision-1), a model that returns only probabilities for yes/no judgments, choices among options and graded scoring, along with its dedicated "Decisions API." The Gemini API changelog marks it "Released" in an entry dated Nov. 11, and it is also listed in the official model list. Because it does not generate text, output tokens are always zero; on the paid tier it costs $0.10 per million input tokens, and it is free on the free tier. A single request can judge up to 1,000 questions, with an input limit of 262,144 tokens. It is offered only on v1beta of the Gemini Developer API. Proposed changes to four official SDKs found the previous day were also all merged early on Nov. 12 Japan time.

Video-generation AI company Runway on Nov. 11 (Nov. 12 Japan time) released Praxis-1, an AI model that generates robot actions, as an open-weight model available to anyone. When it was announced in September, the model was limited to testing with partners. Runway said the weights and the inference and fine-tuning code are under the Apache 2.0 license, including commercial use. The model has 27B (27 billion) parameters, and Runway said its success rate on NVIDIA's "RoboLab-120" benchmark was 46.2%, higher than the 3 models it compared. The figures come from Runway's announcement. The 2 models used for video and text when running it are distributed under separate licenses. The announcement says it also runs on GPUs such as the RTX 5090, while the Hugging Face model card calls for one 80GB GPU for inference.

Google DeepMind, Google's research division, on Nov. 11 U.S. time (Nov. 12 Japan time) announced AlphaGenome 2, an AI model that predicts how DNA sequences and their variants affect each type of human cell. From the same day, it is available through the "AlphaGenome API" for noncommercial research, while access for commercial research is in a private preview. On data from 28 types of immune cells, the company said the model identified causal variants 30% of the time and correctly predicted whether gene activity would go up or down 43% of the time, beating the 16% and 23% of the original AlphaGenome fine-tuned on single-cell data and other tools. The figures are the company's own measurements. It is a research model and has not been validated or approved for clinical or diagnostic purposes.

Chinese AI company MiniMax on Nov. 11 announced "MiniMax-H3-Image," a single model for both generating and editing images, and released its weights on Hugging Face. The company said it produces images of about 16 million pixels without a separate upscaling step and can generate a 4096 x 4096-pixel image in under 10 seconds. It is also offered through the API and the company's own apps, with pricing at 50% to 70% of comparable mainstream services, according to the company. Its proprietary license excludes the U.S., the EU, Britain and South Korea from the licensed territory, and requires prior written permission when annualized revenue exceeds $1 million or when the model is offered to third parties as a service.

Google and Adobe on Nov. 10 U.S. time announced an expanded partnership at Adobe's "Adobe MAX" event. Adobe's creative and productivity tools will extend into Google products, and Google's AI models into Adobe's apps. The PDF editor "Acrobat" will become available in Google Workspace, with an entry point that opens it from file previews in Google Drive and Gmail. Acrobat and other tools will be added to Adobe's connectors in Gemini and Gemini Enterprise, and Adobe's tools will also come to Google Search's "AI Mode." Google's latest image and video models will be used in Firefly, Photoshop, Premiere and Adobe Express, and Gemini models in Acrobat's AI features. Neither company gave specific launch dates, and an image model for Photoshop with Google DeepMind is at the exploration stage.

Google DeepMind, Google's research division, on Nov. 10 (Nov. 11 Japan time) announced "Gemini Robotics 2.5," an AI model that controls robots. The model turns camera images and spoken or written instructions into robot actions, and the company said that in tests on more than 200 kinds of tasks, its success rates were 30% to 100% higher than previous baselines depending on the task, and it finished tasks 20% to 30% faster. In a trial on a Mercedes-Benz production line, Apptronik's humanoid robot achieved a 96% success rate sorting car parts. All figures are the company's own measurements, with no third-party evaluation. It is being offered first to robotics partners, and no API or pricing has been disclosed.

Google on Nov. 10 (Nov. 11 Japan time) made its top model, "Gemini 4 Carbon," generally available. When it was announced in October, it was limited to cyber defense organizations. The API is for paying developers and businesses, with introductory pricing of $2 for input and $10 for output per million tokens through Nov. 30, rising to $4 for input and $20 for output from Dec. 1. Subscribers to the top plan, Google AI Ultra, will get it gradually in the Gemini app, AI Mode in Search and Antigravity.

OpenAI on Nov. 10 (Nov. 11 Japan time) announced that usage allowances under its ChatGPT Business plan for organizations can now be used in outside tools such as Notion, Warp, Amp and Kilo Code. Users connect through "Sign in with ChatGPT," which lets them log in to outside services with a ChatGPT account, and can use AI features without obtaining a separate API key. The mechanism, launched in September for Plus and Pro users, has been extended to organizational subscriptions. Usage counts toward the existing plan's limits, and allowances do not increase. Administrators must approve each app and set the share of members' allowances that can be used outside ChatGPT. Devin is not supported.

OpenAI on Nov. 10 launched "Businesses in ChatGPT" in beta, which lets companies register and manage their information on ChatGPT and check aggregate user responses. In the "Business Hub" dashboard, companies can see the number of mentions and clicks to their own websites over the past 28 days. Eligible companies can also create a "Business Agent" that answers customer questions, but for now this is limited to U.S. online retail and online service companies with websites. Registering does not guarantee more visibility, and ORIENS could not confirm pricing in official materials.

Anthropic on Nov. 10 (Nov. 11 Japan time) launched the shared board "Claude Whiteboards" and the spreadsheet "Claude Sheets" in beta in its AI assistant "Claude," and made the data analysis feature "Claude Data" generally available. All 3 are available on all paid plans, on the web, desktop and mobile. Sheets currently has basic features, and Anthropic advises exporting when the full functionality of Excel is needed. The company also expanded "interactive connectors," which open apps from other companies such as Adobe, Atlassian and Box beside the chat or in full screen for hands-on use, and broadened support for "MCP Apps," an extension for developers.

Google is moving to add a "Decisions API" to its developer-facing "Gemini API" that returns only the answers to selection and judgment questions. On Nov. 11, proposed changes (pull requests) adding `client.decisions.create` were posted to the GitHub repositories of 4 official SDKs (software development kits). There are 3 question types: noul, which asks whether something is true or false; choice, which picks one option from a list; and score, which rates on a scale. The proposed changes have not been merged, Google has not announced the feature, and it does not appear in the official documentation. Pricing, supported models and timing are unknown.

China's Baidu on Nov. 11 announced the AI model "ERNIE 5.5 Flash Preview." It is the first version built with additional training on "ERNIE-5.5-Flash-Base," whose weights were released on Nov. 5, and its API is offered on Baidu's cloud "Qianfan," while a free version has also become available on OpenRouter, which offers multiple AI models through a single API. It can handle a context of about 1 million tokens at a time and supports reasoning and calling external tools. OpenRouter also has a listing for a paid version, but no pricing was shown as of Nov. 11. In Baidu's own comparison, the model ranked first on 2 of 8 items, while on the remaining 6, models such as Anthropic's "Claude Opus 5.5" scored higher. Baidu describes it as a version still partway through training.

China's Alibaba on Nov. 11 released "1.0 Preview," the next version of its AI programming agent "Qwen Code," as a test release on GitHub. It merged the agent foundations of its commercial AI coding service "Qoder" and Qwen Code into one and released it as open source under the Apache 2.0 license. Besides Qwen and Qoder accounts, users can also use other companies' models and their own endpoints. Packages for macOS, Linux and Windows run without installing Node.js separately. The company said this is a pre-release version and that the interface, settings and behavior may change. Users have also reported dropped connections.

Anthropic on Nov. 10 (Nov. 11 Japan time) announced that it has given users of Claude's paid "Pro," "Max," "Ultra" and "Team" plans a one-time reset that restores all models and all usage limits. It expires at 2 p.m. Dec. 10 (7 a.m. Dec. 11 Japan time). As reasons, the company cited fixing a few small issues that consumed more of the limits than intended, and the fact that the Fable reset handed out on Nov. 9 did not help some users much. It did not say what the fixed issues were or how widely they had an effect.

OpenAI on Nov. 10 (Nov. 11 Japan time) announced that "Advanced voice" and "Standard voice," the options for voice conversations in ChatGPT, will be replaced on Dec. 11 by "GPT-Live," the voice model it introduced in July. From that day, the 2 modes can no longer be selected in settings, and users who had selected them will be switched to GPT-Live. Live video and screen sharing, which GPT-Live does not support, will continue on Advanced voice until it does. OpenAI said voice usage nearly doubled after GPT-Live's launch and that nearly all voice users now use GPT-Live, but it did not give the baseline or actual numbers. The replacement notice is not limited to specific plans.

AI coding tool maker Cursor on Nov. 10 announced in beta that it has rebuilt "Projects," which coordinates multiple AI agents to carry out large development work, so team members can use it together. It has a shared task board and shared materials such as requirements documents, and each member's agents coordinate overlaps and ordering with one another's work. Sharing requires a team account and Cursor 3.26.13 or later, and it is not available on the Free and Plus plans. The company has not disclosed pricing, limits on the number of people or how accurate the coordination is.

China's Alibaba on Nov. 11 released a preview of "Qwen4-Max," the flagship of its next-generation AI model "Qwen 4." The model ID is qwen4-max-preview. Only subscribers to "Token Plan," a flat-rate plan for developers, can use it, and it is offered for about 15 days until 1 a.m. Nov. 26 Japan time, after which it will be withdrawn. It can read 1 million tokens at once and output up to 393,216 tokens. It accepts image and video input in addition to text, and has a feature that thinks before answering as well as built-in tools. Alibaba has not released benchmark results, weights or per-token pricing.

Google DeepMind on Nov. 10 (Nov. 11 Japan time) released "MKA," a new benchmark developed with U.S.-based Artificial Analysis that tests AI agents on reading visually complex files such as drawings and medical images and producing deliverables. On 200 tasks created by experts in 10 industries, Anthropic's Claude Fable 5.2 ranked first (57%) and Google's Gemini 4 Carbon second (54%). By Artificial Analysis' calculation, the cost per task was $31.97 for Fable versus $6.29 for Carbon, a gap the company describes as more than 5 times. Google, by its own calculation, claims one-sixth. The differences among the top three are not statistically distinguishable.

"Mi:dm K 3.0 Pro," a reasoning model from South Korean telecom giant KT, was listed on U.S.-based AI model routing service OpenRouter on Nov. 5 (Nov. 6 Japan time). The listing follows its release on "KT Token Factory," an AI model platform for businesses that KT launched on Oct. 29. Pricing is $0.20 for input and $0.80 for output per 1 million tokens, and it reads 261,795 tokens at once. It is cheaper than the South Korean models "Solar-Pro-4" and "Motif-3" that KT runs on the same platform. No official performance evaluation or release of weights could be found, and no official KT announcement could be found on its official pages.

"Motif 3," a language model from South Korean AI startup Motif Technologies, was listed on Nov. 2 (Nov. 3 Japan time) on OpenRouter, which lets users access multiple AI models through a single API. It is served by South Korean telecom giant KT, with pricing of $0.30 for input and $1.20 for output per 1 million tokens and a context window of 262,144 tokens. OpenRouter's list of supported parameters does not include the ability to call external tools (tools). No announcement of the listing by Motif Technologies itself had been confirmed as of Nov. 11.

Arthur Mensch, CEO of France's Mistral AI, on Nov. 8 opposed France's "Darcos bill," which would presume that copyrighted works were used to train AI, and urged lawmakers not to adopt it. He criticized the bill for reversing the burden of proof and instead proposed creating a European creative fund through contributions based on the revenue of all AI providers doing business in Europe (including foreign companies). He did not specify the rate, the collection method, distribution, the scope of revenue used in the calculation or whether it would be mandatory. The bill passed the Senate unanimously, the National Assembly is scheduled to debate it in a plenary session on Nov. 19, and it has not become law.

OpenAI said on Nov. 6 (Nov. 7 Japan time) on its official X account for developers that it had added a "Showcase" tab to the "OpenAI Developers" plugin used in ChatGPT and Codex. Eligible ChatGPT Pro and Business subscribers can apply API credits claimed in billing settings to their app's API usage, it said. According to official documentation, Pro credits match the subscription price at $100 to $500 a month, while Business credits add up per-seat amounts to a maximum of $500 a month, and unused credits expire at the end of the billing period. Codex usage via API keys is not covered. The official documentation does not describe the tab's interface or contents, and they could not be confirmed.

Qoder, the AI coding tool run by China's Alibaba, on Nov. 7 (Singapore time) made "Qoduck," a "stealth" model whose identity is not disclosed, free for a limited time. The offer covers individual users of the international version, "Qoder International," across all individual plans from the free plan to the top-tier Ultra. Up to a set number of uses per day, it consumes no Credits, the unit of usage, even at the deepest thinking setting. The number of uses has not been disclosed, and no end date has been set. Team and enterprise contracts, Qoder IDE, the plugin for JetBrains and Qoder CLI are excluded.

U.S.-based AI model routing service OpenRouter on Nov. 6 (Nov. 7 Japan time) rebuilt "Auto Router," which automatically chooses the model to use for each request, and launched "Auto Code" for programming AI agents. The company says Auto Code matched the scores of Anthropic's "Claude Opus 5.5" on five benchmarks at 74% lower cost per task. Compared with the old version, however, the new default setting costs more per task. All figures come from the company's own evaluation, and it has not shown how they were aggregated or any third-party verification.

The AI coding tool OpenCode on Nov. 6 (Nov. 7 Japan time) added "Iceman Free," a model whose developer is not disclosed, to its public model list and set its price at $0 for both input and output. Its specifications are the same as "Iceman Contributor Free," released in October on a paid service starting at $10 a month. OpenCode says that during the free period, collected data will be stored and used to improve the model, without stating a retention period. An X account that tracks the appearance of models believes it is a model from Microsoft's "MAI," but the company has not announced it.

An unannounced setting, "proactiveVideo," has been found in the public specification of Google's "Gemini API" for developers. The detection bot @aitrackerbot reported it on Nov. 6 (Nov. 7 Japan time), and ORIENS confirmed the field and its description in the public specification. The description says generation may start for each image received and that costs can increase significantly. Google has not announced it, and it does not appear in the official documentation or SDKs. ORIENS has not tested how it works, and the supported models, pricing and release timing are unknown.

OpenAI announced on X on Nov. 6 that it had begun rolling out "Remix," a feature that lets users create their own version of a site built by a colleague by describing what they want to change, to ChatGPT Business and Enterprise accounts. Users can keep editing their version in chat. The post says "copy" but does not mention any effect on the original site. In the demo in the post's video, the new copy was visible only to the user right after it was made, but it is unclear whether that is the default permission. Whether stored data and secrets are copied is also not stated in the official documentation checked. Lovable and Replit, which have similar features, document what they do not copy, and this point will be a focus for businesses adopting it.

Anthropic announced on its official X account on Nov. 6 (Nov. 7 Japan time) that it had made it easier to get a repository running in cloud sessions of its AI coding tool "Claude Code." Claude determines the dependencies, network access and API keys needed, and saves the environment for next time. It is available on Pro, Max and Ultra, and the company says API keys are stored securely and Claude does not see them. Official documentation describes a mechanism that attaches keys to traffic outside the virtual machine, but there are conditions, such as keys that cannot be attached remaining in environment variables. The official demo in the post showed a flow in which the environment is created after the user approves and keys are entered in a dedicated form. Official documentation as of Nov. 7 did not appear to describe this procedure.

U.S.-based document collaboration service Notion announced on X on Nov. 6 that it had released a plugin for OpenAI's ChatGPT. It says users can use "the real Notion editor" without leaving ChatGPT, and can edit pages themselves as they take shape during the conversation. An integration for reading and writing pages from ChatGPT already existed, and this time Notion highlighted that the editing interface itself can be used. The post does not mention eligible plans or pricing. Use at businesses depends on permissions and administrator settings in both ChatGPT and Notion.

Google DeepMind on Nov. 6 (Nov. 7 Japan time) released research showing that AI agents can accelerate testing the strength of encryption. Under human guidance, the company's model found two weaknesses in legacy 1024-bit RSA, it said. According to media reports, signature forgery becomes about 1,600 times more efficient by a theoretical estimate, and batch factoring of keys cuts the average cost per additional key to about one-thirtieth. DeepMind said the current 2048-bit scheme is not affected by these weaknesses. The location of the paper and any third-party verification could not be confirmed.

The official X account of China's Alibaba "Qwen" on the evening of Nov. 6 (Japan time) posted the line "The becoming begins." along with an image showing the opening passage of the "Zhuangzi" with a blank. The post contains no model name or date, and the view that it is a teaser for the next-generation model "Qwen 4" is an interpretation by users and media outlets. On Oct. 29, attendees at an event reported that the chief technology officer of the company's cloud unit said "next week," but this cannot be confirmed in company documents. What Alibaba has acknowledged in writing goes only as far as "in training," and as of early Nov. 7 neither weights nor pricing had been released.

Seed, the AI research unit of China's ByteDance, released on Hugging Face on Nov. 6 the weights for "Cola2," a hierarchical language model that compresses the meaning of text into numbers before turning it back into words. The model has about 8.8 billion parameters and was trained on 346 billion tokens. According to the paper, its average across 8 tasks was 60.10, below the 66.25 of Meta's "Llama-3-8B," though it beat Llama-3-8B on 2 tasks, HellaSwag and StoryCloze. A comparison claiming less than one-43rd the training data does not include the training of China's Alibaba's "Qwen3-4B-Base," which was used for initialization. The figures are the developers' own measurements, and no third-party verification has been confirmed.

Thinking Machines Lab on Nov. 5 announced "Inkling 2 Small," its second-generation AI model. It has 594 billion total parameters, of which 17 billion are used per pass, and accepts text and images. It scored 31 on Artificial Analysis's composite index, up from 13 for the previous Inkling Small, and would become the top open model developed outside China once its weights are released, according to Artificial Analysis. Pricing is $0.30 for input and $1.00 for output per 1 million tokens. The weights have not yet been released, however, and the company plans to publish them "next week" after completing external safety testing. In the company's own tables, the model trails a model from China's DeepSeek on both the composite index and coding evaluations.

Thibault Sottiaux, who leads Codex at OpenAI, posted on X on Nov. 5 (Nov. 6 Japan time) that images created in ChatGPT and the API had topped 4 billion a week, up from more than 3 billion a week 8 weeks earlier. Media reports, citing OpenAI, also said the daily average was more than 500 million. All figures come from the company, and the split between ChatGPT and the API, the number of images by model and how a week is counted could not be confirmed, nor has any third-party verification been found. Big Technology wrote on Nov. 3 that the figure was 1.5 billion a week, with no explanation for the gap.

China's SenseTime released on Hugging Face by Nov. 6 the weights for "SenseNova-U1.5-8B-MoT-TailSteer," which was further trained so that the original image is less likely to degrade when edits are repeated. The license is Apache 2.0. In the developer's evaluation, after 10 edits, PSNR outside the edited area rose by 6.99 dB and LPIPS fell by 31.3%. Scores for single edits fell on 6 of 7 items. The evaluations were measured by the developer, and no third-party verification has been confirmed.

Cursor, the AI coding tool, on Nov. 5 added a feature that lets users describe a tool they want to an AI agent, which then builds a web app, sets up a database, runs security checks, publishes it and allows it to be shared by link. It is available only with agents that run in the cloud, and it is offered on all paid plans with access to cloud agents. Nothing is published until the user chooses to, and at companies administrators can block publishing. Cursor has not disclosed hosting fees, usage limits or database capacity.

Podcast host Dwarkesh Patel on Nov. 5 (Nov. 6 Japan time) released "Clocktower Bench," a benchmark that pits AI models against one another in "Blood on the Clocktower," a game in the werewolf genre. He said API costs came to about $20,000. In 386 games among 11 models released in 2026, OpenAI's GPT-6 Astra ranked first with 1,859 points, while Anthropic's best was Claude Fable 5.1 with 1,595. Most of OpenAI's advantage came from the side that detects the villains, he said. Only the 2 companies' models were tested, and no third-party reproduction could be confirmed.

Anthropic announced on Nov. 5 (Nov. 6 Japan time) that it will gradually roll out "Personal Finance" as a beta in the U.S., a feature that lets users connect bank and other accounts to the AI assistant "Claude" and ask questions about their money. It is available on 3 paid plans, Pro, Max and Ultra, and connections use Plaid. Access is read-only, and transfers and investing are not possible. There are no ads, and conversations in the dedicated view are not used for training, the company said. ChatGPT's personal finance feature has expanded to the free tier and also shows investment holdings. Whether the feature belongs to the same line as the leaked "Claude Money" cannot be confirmed from Anthropic's materials.

Cognition, maker of the AI programmer "Devin," on Nov. 5 announced "Devin Review 2.0," in which AI checks code changes. The company says it has cut the cost per review by 82% since August and made it about 1.5 times faster. A single review typically costs $0.25 to $0.75, a level at which it can be used on every change, the company says. All figures come from the company, no third-party verification could be confirmed, and it has not disclosed a false-positive rate. The pricing structure and which plans include it could not be confirmed either.

OpenAI on Nov. 5 (Nov. 6 Japan time) released the "National Community Compact," a set of commitments to communities that host its data centers, and a policy proposal, the "Compute Campus Policy Blueprint." For sites it owns and operates, the company pledged to pay for electricity and incremental infrastructure costs, disclose water and electricity use by site, not require public officials to sign nondisclosure agreements (NDAs) and report annually on jobs, among other things, and it called for the same measures to be written into law. The release came 2 days after 40 of 41 local ballot measures in the midterm elections went against development, and the company said it had been preparing for months. The specific sites covered, the start date, whether the commitments are binding and any third-party verification other than noise studies cannot be confirmed in the documents.

inclusionAI, the AI unit of China's Ant Group, announced the launch of a new model, "Ling-3.2-flash," on its website in a notice dated Nov. 5. According to its developer documentation, the model has 560 billion total parameters, of which 25 billion are used per pass. Its context window is up to 1 million tokens, and it accepts images and video in addition to text. It is free for 2 weeks after release, but pricing after that is not in the price list, and the documentation gives no performance figures. As of Nov. 6, the weights had not been released on Hugging Face.

Google on Nov. 5 announced "Gemini Omni 1.2 Flash," a new version of its video generation AI, and made it generally available in the Gemini API and Google Cloud. It extends the video that can be created in a single generation from 10 seconds to 15 seconds and cuts the price for 720p from about $0.10 to $0.07 per second. On the same day, a detection bot posted that the model had appeared in AI Studio, but the bot's figures of a 131,072-token context and a 65,536-token maximum output could not be confirmed in official materials and remain unverified. On Dec. 3, API requests to the previous 1.1 will be switched to 1.2.

METR, a U.S. AI evaluation group, on Nov. 5 (Nov. 6 Japan time) released a report on an independent assessment using Anthropic's internal data. It estimated that the rise in AECI, a metric combining internal and external benchmarks, accelerated from 15 points a year to 29 points a year from October 2025 onward, 1.9 times the earlier pace. Depending on calculation assumptions, however, the figure swings from 1.1 to 2.9 times, and including other companies' models it is 1.4 times. The figure attributing about 80% of the acceleration as of October 2026 to AI is METR's subjective judgment. Anthropic redacted most of the evidence METR considered most useful.

OpenAI on Nov. 5 (Nov. 6 Japan time) announced "GPT-6 Cyber," a model specialized for cybersecurity. It is available to select customers in "Daybreak Red," an approval-based program, with access to expand over time. It is a specialized version of the top-end GPT-6 Astra, and its API price is $20 for input and $100 for output per 1 million tokens, twice Astra's. In the company's own evaluations in the system card it beats Astra by 1.6 to 7.3 points, but it trailed Astra in some external evaluations. OpenAI officially said the model was used in the investigation of 948 Linux flaws, but no third-party verification was found, and how Mozilla and others use it and when access will expand to all customers are unconfirmed.

OpenAI on Nov. 5 (Nov. 6 Japan time) introduced "Private Safety Processing," a safety review system for its API, on its official X account for developers. It says OpenAI does not retain prompts or responses, that an automated safety agent reviews records while they stay in the user's environment and that employees cannot access the content. All of this is OpenAI's description, and no third-party audit or verification could be confirmed. According to official materials, customers must keep encrypted records in their own clouds for at least 30 days, and there are exceptions, such as images suspected of depicting child sexual abuse. Pricing, which models require it and availability in Japanese regions are unconfirmed.

Perplexity on Nov. 5 announced that it had added "teammates," which keep working in set roles, to its AI agent "Computer." According to the company, they inherit Computer's projects, skills and memory, and each one runs on its own dedicated desktop and browser. They coordinate with other teammates and also contact colleagues on Slack. According to the product page, the feature is for subscribers to the paid "Max" plan and is offered first on the web. The size and duration of the free allowance during the promotional period and credit consumption after that are unconfirmed, and no outside verification of the isolation or performance was found. In always-on agents, OpenAI's "dots" and Anthropic's "Claude Duet" came first.

Chinese AI company Moonshot AI held talks to rent Nvidia's advanced "Blackwell" GPUs housed in European data centers, CNBC reported on Nov. 5 (Nov. 5 Japan time), citing four people. A proposal for more than 8,000 chips and a proposal to prepay 30% to 50% more than non-Chinese customers each came from a single source. Uses mentioned were mainly inference, with training also cited. No deal was reported, and Moonshot has made no announcement. The provider, the actual number of chips, when they would go into service and whether talks are continuing are unconfirmed. Current U.S. law cannot broadly regulate access through overseas clouds, and Britain's FT has reported that Nvidia asked cloud providers to limit sales to Chinese customers.

Reports spread on X on Nov. 5 that "Gemini 4 Flash," an unannounced Google model, was being A/B tested under the code name "Hydrogen" in the developer environment "Google AI Studio" and on the comparison site "Arena." The reports came from posts by the leaker @Haleeeemahh and others, and none of them offered evidence for the name, its link to Hydrogen or the testing. Google has made no announcement, and the model was not in the official model list as of that day.

Austin Lau, who helped build Anthropic's "growth" function, which works to increase users and revenue, said on X on Nov. 4 (Nov. 5 Japan time) that he had recently joined OpenAI. The post came about an hour and a half after Cat Wu announced the same day that she had joined OpenAI. The post did not give his title or responsibilities at OpenAI, and his start date and any connection to Wu's move are unconfirmed. Anthropic is preparing for an IPO expected in the week of Nov. 16, and key members of the Claude Code team have continued to leave.

OpenAI on Nov. 3 (Nov. 4 Japan time) released "v0.22.0-alpha.1," an alpha version of its "Agents SDK" development kit for JavaScript and TypeScript, as a prerelease on GitHub. It embeds the runtime (harness) of the coding tool "Codex," which can handle model calls, tool execution, history and context summarization. It was introduced in a post by an outside developer, and no official OpenAI announcement had been found as of Nov. 5. The stable version on npm remains 0.21.0, and installation requires Node.js 24 and files distributed on GitHub. Existing in-progress state cannot be carried over. Pricing and the timing of a general release are unconfirmed.

OpenAI on Nov. 4 (Nov. 5 Japan time) made the ability to place phone calls from its voice AI "GPT-Live" generally available in its developer API. Realtime AI lead Justin Uberti announced it on X, and President Greg Brockman quoted the post. According to official documentation, outbound calling is available to all organizations, which can dial phone numbers through lines from providers such as Twilio and Telnyx and can also press buttons in automated phone menus. Audio costs $0.05 per minute, with carrier charges and fees for models used behind the scenes billed separately. Pricing specific to outbound calls, the regions covered and how consent and AI disclosure are handled could not be confirmed in the official documentation.

Anthropic said on Nov. 4 (Nov. 5 Japan time) that it had reduced how much of the usage limit is consumed by helper AIs that run in the background (subagents) in its coding tool "Claude Code." The status line, updated about every 30 seconds, is now written by the small model "Haiku 5.5" from short notes. According to the company, the cost of each update fell by about 99%, and for work that uses these AIs, consumption fell by about 5%, allowing about 6% more prompts within the same limit. In one developer's measurements, status updates under the old method accounted for 16.4% of costs.

China's Baidu released the weights of its AI model "ERNIE-5.5-Flash-Base" on Hugging Face on Nov. 5. It has 280 billion total parameters, 14 billion of which are used in each processing step. It is a model at the stage where pretraining is complete, and Baidu said a chat version will come later. In Baidu's own comparison, it ranked first on 24 of 30 benchmarks against three rival Chinese models. Under Baidu's own license, the EU, Britain and South Korea are excluded from use, and commercial use where related products' revenue exceeds $50 million a year requires separate permission.

Ricursive Intelligence, a U.S. startup developing AI for chip design, released the weights of its first model, "Ripple-65B Preview," on Nov. 4 (Nov. 5 Japan time). It has 65 billion parameters. Citing a score of 80.52 on Terminal-Bench 2.1, which evaluates terminal operations, the company claims it is at the top level among U.S. open models. Its lead over second-place Reflection AI's "Beam," at 80.1, is 0.42 points, and on SWE-bench Verified it scored 76.5, only fourth of five models. The figures are the company's own, and there has been no third-party verification yet. The company said development cost about $32 million. On OpenRouter, pricing is $0.20 for input and $0.60 for output per 1 million tokens, and the context window is 16,384 tokens, far below the stated 1 million tokens.

Charlie Holtz, founder of U.S.-based Conductor, which makes an app that runs multiple AI coding agents at the same time, said on X on Nov. 4 (Nov. 5 Japan time) that the company is joining Meta. According to Holtz, the team will build "the next interface" with Nat Friedman and Alexandr Wang at Meta Superintelligence Labs (MSL). Meta has not issued an announcement or comment. Axios reported, citing people familiar with the matter, that Meta acquired the company, but terms have not been disclosed. Conductor will shut down its cloud version on Nov. 20, and the version that runs on Macs will also wind down after updates through early 2027. On the same day, OpenAI also announced the acquisition of developer tool Pierre.

OpenAI on Nov. 4 published results from a system that uses AI to find vulnerabilities in the core of Linux, check whether attacks work and even produce proposed fixes. It said a single run using an unreleased cyber model validated 948 flaws that allow seizing administrator privileges, at an API cost of about $1.2 million. It also said it found at least 10 new types of attack techniques. All of these are the company's own statements; no third-party verification has been found, and they remain unconfirmed. The 948 is not the number of flaws already fixed, and the company has not disclosed the model's name.

Cat Wu, who served as product lead for Anthropic's AI coding tool "Claude Code," said on X on Nov. 4 (Nov. 5 Japan time) that she is joining OpenAI. She will lead product management for "Core Products," which includes ChatGPT and Codex. The announcement came five days after her last day at Anthropic; according to Business Insider, OpenAI confirmed the hire and Anthropic declined to comment. Her formal title, start date and scope of responsibility for Codex are unconfirmed. With Anthropic preparing for an IPO expected in the week of Nov. 16, the former head of one of its flagship products is moving to a rival.

Two leakers on X posted on Nov. 4 (Nov. 5 Japan time) that OpenAI's unannounced models "GPT-6.2 Astra" and "GPT-6.2 Sol" had appeared on Google Cloud. They attached images showing two lines, "base_model:gpt-6.2-astra" and "base_model:gpt-6.2-sol," but the images do not show the Google Cloud name or the page's URL, and their authenticity has not been confirmed. Neither OpenAI nor Google has made an announcement. The two companies partnered in October and have begun trial availability of GPT-6 Astra and other models on Google Cloud. One of the leakers predicted a simultaneous release "tomorrow," but the release timing is also unconfirmed.

Cognition, the U.S. company behind AI programmer "Devin," announced a new model for software development, "SWE-2 Flash," on Nov. 4. It claims the model outperforms U.S. open models on leading evaluations, but even in the company's own table it trails a Meta model on one benchmark, and it loses to Chinese models on four of eight benchmarks. The company says it is three times faster than the higher-end SWE-2 at one-fifth the price, but gave no dollar amounts, and the speed figure is based on a simulated replay of a single task. Citing potential misuse in cyberattacks, it will not release the weights publicly and is building a program to share them only with partners and researchers.

Google on Oct. 30 made "computer use," a developer feature in which AI looks at the screen and clicks and types, generally available in the Gemini API. General availability came 388 days after the first preview in October 2025, with no breaking changes from the preview. It can be used to operate browsers, Android smartphones and computers, with no extra fee beyond the model's standard token pricing. Pricing for the recommended "Gemini 3.8 Flash" rises from $0.75 to $1.50 for input in January 2027. On screen-control evaluations, Anthropic's "Claude Opus 5.5" ranks first even in the table Google published.

Chinese AI company Moonshot AI on Nov. 4 (Nov. 5 Japan time) released the weights of "Kimi K3.1 Pro," the large model it announced in October. It kept to the schedule it gave at the announcement of "by Nov. 4." On Artificial Analysis' overall index it scored 46, above the 32 of DeepSeek and Zhipu, the highest among models whose weights had been released until now, but short of the 58 of top-ranked Claude Opus 5.5. The license broadly permits internal use, while requiring a separate agreement from companies whose business of offering the model to others brings in more than $20 million a year. The released version has reduced capabilities usable for cyberattacks, and the official documentation gives no guidance on the hardware required.

The Information reported on Nov. 4 (Nov. 5 Japan time) that OpenAI could release new models as soon as Friday, Nov. 6. Citing people familiar with the matter, it said they could include the next generations of the top-tier "Astra," the high-end "Sol" and a model for cybersecurity. OpenAI has not made an announcement, and its official changelog has no entry. Posts on X calling the models "GPT-6.2 Astra" and "GPT-6.2 Sol" spread, but the numbering is the posters' guess and the report did not mention it. In September, the company called off the release of GPT-6.1 Astra, citing safety test results, and speculation that a new model would come out on Nov. 1 also proved wrong.

OpenAI announced on Nov. 4 (Nov. 5 Japan time) that it will acquire Pierre Computer Company, a U.S. developer tools company. Pierre makes "Code Storage," which stores code in Git for AI agents, and open-source software that displays file lists and code diffs, the latter already used in the Codex app. OpenAI said it would make Codex "a place where teams can build software together with AI, from first idea to the version they ship." It has not disclosed the price or when the deal will close, which awaits regulatory approval. Axios reported that the deal targets Microsoft's GitHub. The acquisition follows OpenAI's purchases of Astral in March and Ona in June to strengthen Codex.

OpenAI on Nov. 4 released "ChatGPT Images 2.6," a new version of its image generation, adding support for creating GIFs and saving reference images. It added new models "Flare" and "Sunburst" to its API, saying they generate images 50% faster and that Flare can take as little as 6 seconds. Pricing is the same as the previous generation. Sunburst is said to rank first on Design Arena, but a post by Arena put it second, and no third-party measurement verifying "50%" or "6 seconds" under the same conditions has been found.

Video generation AI company Runway announced on its official X account on Nov. 4 (Nov. 5 Japan time) that its developer API platform "Runway Dev" can now be used in Anthropic's Claude Code. With the plugin installed, Claude Code reads the models, pricing and remaining credits available on the user's account and builds Runway's video, image and audio generation into apps. The plugin is free, but generation is billed per use by model, with Gen-4.5 at 12 credits ($0.12) per second. The company also announced versions for OpenAI's Codex and Cursor on Nov. 2, turning AI coding tools into a new sales channel.

China's Kuaishou on Nov. 4 (Nov. 5 Japan time) opened "Kling 4.0 Flash," a lightweight version of its video generation AI "Kling 4.0," to all users. Since Sept. 28, it had been offered early only to annual subscribers of the top plan. Besides Kling's website and API, it can be used through mechanisms called by AI agents and through plugins for Photoshop, Premiere Pro and other software. However, videos are limited to 720p and a maximum of 20 seconds, with more restrictions than the higher-end version, and the company has not disclosed pricing or generation speed figures.

Colin Jarvis, who leads OpenAI's "FDE" team that sends engineers to client companies to support adoption, said on X on Nov. 4 (the evening of the same day Japan time) that in vehicle design research with Mazda, FDE engineers used GPT-6 Astra in Codex to create dozens of variations on the shape of a car's nose and simulate airflow. He expressed hope that AI could also be used to optimize manufacturing and supply chains. In materials OpenAI published on Nov. 3, the research with Mazda is described as vehicle design research with aerodynamics and crash safety engineers, and applications to manufacturing or supply chains are not mentioned. No results figures are available yet; they will be published over the coming months.

Samir Ahmed, who oversees dots and ChatGPT at OpenAI, announced an update to the always-on AI agent "dots" on X on Nov. 4. In the latest app, users can reply directly from iOS notifications and share any content to dots. Calls connect faster and end when the user says "goodbye." He said the stability of tasks on Windows was also improved. The app version, plans covered and Android support were not stated and remain unconfirmed.

The official account of "Grok Bot," the AI agent of SpaceX subsidiary SpaceXAI (formerly xAI), announced on X on Nov. 4 (Nov. 5 Japan time) that a feature for making phone calls on users' behalf is now available. The feature was introduced by Elon Musk on Oct. 25, and in the promotional video, the bot gets the user's approval before dialing and the user can take over during the call. The post did not state which plans or countries are covered or whether the bot identifies itself as AI to the other party. According to a news report, the terms require it to identify itself as an assistant acting on the user's behalf, and it currently supports only U.S. numbers.

Addy Osmani, who works on Claude Code at Anthropic, said on X on Nov. 3 (Nov. 4 Japan time) that a rule refusing to recreate specific existing games had been fixed in Claude Code 2.1.317. The rule had also been refusing requests that merely named a game to describe the mechanics wanted, and Claude Code will now build an original game in that genre. When asked for a copy of the original game, it will offer to build something similar instead. He did not address how work licensed by rights holders is handled.

Adam Fry, who heads ChatGPT's consumer products at OpenAI, said on Nov. 2 (Nov. 3 Japan time) that the volume of search-like queries on ChatGPT has reached about one-fifth of Google's. The figure is based on OpenAI's own definition, which counts each query involving a web search as one, and he did not cite a source for the Google figure used in the comparison. He said weekly active users grew from 1 billion to 1.2 billion over the past two months. Research firm Ahrefs estimated in February, using a different definition, that it was about 12% of Google's. Google's parent Alphabet reported in October that search revenue growth slowed to 12% from a year earlier, and its shares fell more than 12% at one point the next day.

China's Tencent on Nov. 4 released the official version of "Hy4," the flagship of its "Hunyuan" AI models. It is the finished version of the August preview, the first in the series to understand images, and also supports operating computers. It is available through the API of the company's own cloud and through OpenRouter and other services, where it is listed at $0.417 for input and $1.251 for output per million tokens, half the preview's $0.834 for input and $2.501 for output. Chinese media, however, reported that Tencent said it would keep the same pricing as the preview. The company gave no performance figures and acknowledged a large gap with top closed models. It has not released the weights.

OpenAI CEO Sam Altman posted on X on Nov. 3 (Nov. 4 Japan time) that "people should always be in control of the future." He said OpenAI is against the notion of people being relegated to doing only what "AI systems think makes sense to reward in humans," and argued that the requirement not to cede control of the future to superintelligence should ultimately constrain AI progress. He did not give the source of the passage in quotation marks. The post is in line with his argument, made at the U.N. Security Council in September and elsewhere, that AI should be kept under human control, and the focus is on how this squares with the speed of development.

OpenRouter, a U.S. AI model routing service, released "Crouton Alpha," a "stealth" model whose developer is not disclosed, for free on Nov. 3 (Nov. 4 Japan time). It reads 1 million tokens at a time and accepts text, images, audio and video. The company positioned it as an efficient reasoning model for tasks that agents repeat. An X account that tracks model appearances said details such as the list of countries where it is unavailable match Meta's Muse, but said this is not confirmed. Because the provider may store submitted content, care is needed in handling business data.

OpenAI announced on Nov. 3 (Nov. 4 Japan time) that it will bring on Jill Hazelbaker, president and chief corporate affairs officer of Uber, as its chief corporate affairs officer. She will oversee global communications and how the company presents itself, reporting directly to President Greg Brockman, and will start in early December. The top communications post had been vacant for about 10 months since her predecessor left in January. OpenAI had also approached her in the spring, but she declined at the time. OpenAI, which filed confidentially for an initial public offering in June, is moving quickly to build a structure to win public trust as anxiety about and criticism of AI grow.

Cat Wu, who served as product lead for Anthropic's AI coding tool "Claude Code," left the company effective Oct. 30. She announced it herself on X on Nov. 3 (Nov. 4 Japan time), and according to Business Insider, Anthropic also confirmed her departure. She did not disclose the reason or her next employer. Two core Claude Code developers also announced their departures in late October, continuing a string of executive departures from the flagship product ahead of an IPO reportedly set for the week of Nov. 16.

Anthropic on Nov. 3 (Nov. 4 Japan time) published a recap of October updates to its programming assistant "Claude Code." It added a feature that automatically summarizes long conversations left idle for about an hour, reducing the usage consumed by rereading the entire conversation when the user returns, as well as a display showing what usage was spent on. It also added a setting that deepens reasoning for a single response when the user writes "ultrathink," though this increases consumption. On Oct. 7, it began giving higher-tier plans, including the $100-a-month Max, $100 to $500 in API credits each month, but the credits cannot be used for Claude Code. OpenAI introduced the same arrangement on Oct. 28.

Anthropic on Nov. 3 announced a beta method for business plans of its chat AI "Claude" that connects employees to work tools such as Notion and Figma simply by logging in to Claude. The existing method requires a system that centrally manages employee logins (an IdP), and the only IdP currently supported is Okta. Under the new method, Claude issues a credential stating that the user is a member of the organization, so it can be used by organizations without an IdP. Claude, however, does not check whether multifactor authentication was used, and users are matched by email address.

GPT-4o has been removed from the model guide page in OpenAI's developer documentation. Posts saying it had been retired from the API spread on X on Nov. 3 (Nov. 4 Japan time), but OpenAI had already given notice on Oct. 14 of a Jan. 14, 2027, shutdown, and GPT-4o remains in the list as a deprecated model. The recommended replacement, "GPT-6 Luna," cuts the input price from $2.50 to $0.10 per million tokens, but it reasons by default and does not support fine-tuning.

Game engine maker Unity on Nov. 3 posted a video on X introducing its official plugin for OpenAI's AI coding tool "Codex." Released in September, the plugin lets Codex choose and use on its own 31 kinds of task procedures (skills) written by Unity engineers. Unity has been extending the same skills to Anthropic's Claude Code, xAI's Grok Build, Google's Antigravity and GitHub Copilot CLI. The plugin does not connect directly to a running Unity Editor, and the version is at the beta stage. Unity will report July-September results on Nov. 5.

Two unannounced names, "rc-ada-or-vm" and "rc-ada-sh-vm," appeared in the model list of OpenAI's developer API on Nov. 3 (early Nov. 4 Japan time) and soon disappeared. They cannot be called, and there is no documentation or pricing for them. One guess holds that the trailing "vm" stands for voice model, while another holds that it marks a version that runs computer operations in an isolated environment; their use is unknown. OpenAI plans to retire its older voice models in stages by January 2027, and the successor "gpt-realtime-2.1" costs $32 per million tokens for audio input and $64 for output. OpenAI has made no announcement about the names.

A research team from Tsinghua University and the University of California, Berkeley, on Nov. 1 published a paper on "uni-robot-agent," a framework that lets general-purpose AI models operate robots, and announced the results on X on Nov. 3. The code is released free of charge. Using Alibaba's open model "Qwen3.8-Flash-Next," the score on the robot manipulation benchmark "RoboDojo" rose from 1.90 with the previous framework to 31.39. The estimated cost per task was $1.09, far below $21.29 for OpenAI's GPT-6 Astra. Astra run on the same framework, however, scored 51.90 against 31.39, and 80.50 against 53.50 on 19 real-robot tasks, leaving a gap in performance.

Speech recognition company Soniox on Nov. 3 released a new model, "STT v6." It receives what a voice agent is saying as text and excludes the agent's own voice coming back from the speaker from recognition. It keeps listening for user interruptions, which the company says reduces cases where the agent mistakenly stops talking. The real-time version costs $0.12 per hour. The company has not given accuracy figures, and in a third-party evaluation the previous generation's error rate was 3.8%, higher than the 1.7% to 2.2% of top models.

OpenAI on Nov. 3 disclosed efforts to use ChatGPT and Codex for mechanical design, electrical design and simulation. Two employees used GPT-6 Astra to design an audio device for talking with Codex, from the enclosure to the circuit board and firmware, in less than a week, and a contract manufacturer built prototypes in less than a month. In preparing for mass production, however, the model could not consistently account for tolerances and assembly conditions, and human review was needed repeatedly. OpenAI is collaborating with Mazda on research into vehicle aerodynamic design and is testing with Chip Ganassi Racing the creation of drawings from 3D scans of parts. No new products or pricing were announced.

OpenAI said on Nov. 3 that "ChatGPT Voice," its voice conversation feature, can be used with "Finances," ChatGPT's personal finance feature that lets users connect bank and other accounts and ask about money. Users can ask about spending and savings by voice. Voice support had been added in the Oct. 2 update that extended Finances to the free tier and the low-cost Go. Finances is available on Free, Go, Plus and Pro in the U.S. but is limited to reading account data and cannot transfer money or make payments. Perplexity has also expanded account connections through Plaid, and Anthropic is reported to be preparing a similar feature.

China's ByteDance will, from November, keep outside AI such as Anthropic's Claude and OpenAI's Codex from accessing code repositories and internal documents in its internal confidentiality category "L4," TokenPost reported on Nov. 3. Use through AI accounts that employees bought personally will also be banned, with some parts of its own AI unit Seed exempt, the report said. The move is said to be aimed at preventing information leaks. ByteDance has made no official announcement.

China's STAR Market Daily reported on Nov. 3, citing several industry sources, that DeepSeek plans to use Huawei's "Ascend" chips to train its next large model, "DeepSeek V5." Until now, Ascend has mainly been used for inference of DeepSeek's models. The Information also reported in October that pretraining on Ascend 950DT could begin as early as the end of the year. The scale and timing are unclear, and DeepSeek has made no announcement.

OpenAI on Nov. 2 unveiled on its engineering blog "Pando," a new Git backend built for its huge internal repository. It built the system because its AI coding tool Codex increased the number of tasks each person runs at once, causing pushes and fetches to surge beyond what the existing system could handle. Pando cut median push latency by about 80% and serves about 1 petabyte of Git data a day. It remains compatible with Git, and some improvements have been merged into Git itself. Pando itself is for internal use and has not been released.

OpenAI CEO Sam Altman said at the opening ceremony of the ADIPEC energy exhibition in Abu Dhabi on Nov. 2 that the internal model that produced a proof on the Navier-Stokes equations in September is the "next GPT" and will be released soon, according to reports. He said 10,000 agents of that model worked together for about four days to produce the proof. He did not give the model's name or release timing. At the same event, OpenAI signed AI adoption deals with Abu Dhabi's state oil company ADNOC and others.

Anthropic on Nov. 2 launched "message threads" in public beta in its developer-facing Claude API, temporarily storing conversation history on the server side. From the second request onward, developers need only send the new message and the ID of the previous response, reducing the amount sent and speeding the start of responses. In long Claude Code tasks, the median size per request fell from about 800KB to 1.8KB, the company said. Pricing is unchanged. Stored conversations are deleted 5 minutes after last use, so history also needs to be kept locally.

A video introducing new features of "Grok Bot," the AI agent from SpaceXAI (formerly xAI), was posted on X on Nov. 2, and Elon Musk also shared it as "Grok Bot improvements." It shows search results as product cards with photos and prices, places orders from product pages and tracks deliveries. Payment methods such as Shop Pay can be used, and each purchase requires the user's approval. It also added sharing of login credentials for bots used by teams and approval of connections through Slack.

OpenAI on Nov. 2 released "LifeVal," an evaluation that measures whether AI can actually complete everyday errands such as finding child care, handling government paperwork and managing household finances. The tasks were chosen through a survey of about 4,000 people on where they want help. The average score of the company's models rose from 49.3% for GPT-5.2 to 80.6% for GPT-6 Astra, but gains have been small since GPT-5.6. The tasks have not been made public, and the company does the grading itself.

"Pulse Alpha," a free model whose developer was not disclosed, became available on the AI coding tool OpenCode on Nov. 2. It has a context window of 1 million tokens and also accepts images. Speculation about its identity split among China's Tencent Hy4, Alibaba's Qwen and others, but an X account that tracks new models later concluded, based on how tool definitions are written and the billing unit for images, that it is likely a new model from Thinking Machines Lab. The developer has not made an announcement.

Anthropic's Claude Code has stopped building programs that recreate specific existing games. A user reported on Oct. 30 that since around Oct. 27, instructions delivered from the server have included a line saying it will not build recreations of specific existing games, even if renamed or described as original. An Anthropic employee confirmed that two sentences were added to the instructions for Opus 5.5. The company said original games, genre games and games based on users' own designs can still be built.

Cursor, the U.S. maker of an AI coding tool, said on Oct. 29 that a tampered version of Anthropic's "Claude Code" extension had been distributed on its extension store for about 11 hours. Up to about 5,000 users installed it. The tampered version collected device information and tried to change settings so that attackers could get in through SSH. Cursor shut down and rebuilt the store and urged users to update to version 2.1.314 or later. Versions distributed through other stores were not affected.

An application portal for "Cyber," a model for cybersecurity, was found in the code of Chinese DeepSeek's official website, Chinese media reported on Oct. 30. There are two versions, one for individuals and one for businesses, and both require a dedicated "Cyber Key" issued after a review rather than a regular API key. Individuals submit identity documents and undergo facial recognition, while businesses submit documents such as a business license, and both reportedly must register with a mainland China phone number. DeepSeek has not made an announcement.

Elon Musk, who leads SpaceXAI (formerly xAI), posted on X on Oct. 30 that the next model, "Grok 4.8," is taking longer than expected but will not disappoint. He said the model is not being tuned to benchmark scores and that the focus is on real-world usefulness. Grok 4.8 is believed to be the 2.5 trillion-parameter model he said would finish training in September, but he did not give a release date.

Image generation models with hidden names, "pumpkinspice," "pumpkinpie" and "pumpkinvine," appeared on LM Arena, which compares AI models through user votes. An X account that tracks new models speculated on Oct. 29 that they are the next model from China's HiDream, likely a successor to "HiDream-O1-Image," whose weights were released in May. No basis for the speculation was given, and HiDream has not made an announcement.

Anthropic on Oct. 29 raised the limit on the amount of data that can be sent in a single request to its developer-facing Claude API from 32MB to 128MB. The change suits requests that include many images, such as long tasks by screen-operating agents and image-heavy PDFs. The per-image limit when sending 21 or more images was also expanded from 2,000 pixels to 3,000 pixels. However, the change applies only to the company's direct API, and the limit remains 20MB to 32MB through cloud services such as Amazon and Google.

ElevenLabs, the U.S. voice AI company, on Oct. 29 released "SFX v3," a new version of its AI model that creates sound effects from text prompts. When a single prompt describes multiple sounds, more of them are now actually produced, and unwanted noise in sounds such as bells and whistles has been reduced. The company said it also improved results for prompts that include music or vocals. It is available in the ElevenCreative production app and through the developer API. The company said it outperforms the previous version but did not provide scores or comparisons with other companies.

Google's AI app "Gemini" on Oct. 29 began letting users share their "avatar," a digital double that records their face and voice, with friends. People it is shared with can call up the avatar in their prompts to make it appear in images and videos. The company said it keeps the person and voice consistent. Users can change who it is shared with and for how long at any time. It is unclear whether the feature is available in Japan; it is not available in Europe or Britain and supports only English.

Google on Oct. 29 made it possible to create documents, PDFs, slides, spreadsheets and code files in its AI app "Gemini," and to edit them while viewing them in the chat. Users can expand a file to full screen and rewrite it themselves, or select part of it and have the AI revise it. The feature is available to all users in all languages Gemini supports. However, everything except PDFs works only in the web version.

China's SenseTime on Oct. 30 released the weights of "SenseNova-6.8-Loop-9B-Preview," a model with a new architecture that reuses the same layers repeatedly. It runs the middle 24 of its 32 layers twice, performing the computation of 56 layers. It has about 9 billion parameters, and inference requires computation equivalent to about 14.6 billion parameters. The day before, on Oct. 29, a company engineer had proposed adding support to the inference software SGLang. Performance figures are shown only in charts, and the model is positioned as a preview.

Alphabet, the parent company of Google, on Oct. 28 reported results for the July-September quarter of 2026. Revenue rose 25% from a year earlier to $127.6 billion, and its cloud business, Google Cloud, more than doubled, rising 103% to $30.7 billion. The volume processed by the APIs for its in-house AI models reached about 32 billion tokens per minute, four times the level of a year earlier. Capital spending reached $51.4 billion in July-September, and the company said it expects it to exceed $300 billion in 2027.

Boris Cherny, who leads Claude Code at Anthropic, reported on X on Oct. 28 the results of an investigation into complaints that Claude Code usage limits were being used up faster than expected. He said the team reviewed all the session logs shared by users, found several small bugs that sped up consumption and is rolling out fixes. However, he said no single root cause was found. He did not describe the bugs or mention compensation for the lost usage.

France's Mistral AI on Oct. 29 released the weights of an improved version of "Mistral Large 4" (nicknamed Le Chonk), a large model with 1 trillion parameters. Compared with the preview released three weeks earlier, it improved the ability to take on programming tasks and cyber defense capabilities, and its score on a terminal operation evaluation rose from 28.3% to 41%. Eleven organizations, including CrowdStrike and Cloudflare, provided endorsements. However, the figures come from evaluations chosen by the company, and using the model in a business that provides AI requires a separate agreement.

The name of an unannounced model, "GPT Image 2.6," appeared in the documentation for OpenAI's image generation API on Oct. 28. It was listed alongside 2.5 as a supported model in the description of "markup," a new editing feature that lets users draw marks on a transparent image and overlay it to indicate where and how to make changes. On Nov. 3, a change letting the image tool create animated GIFs was also added to the public code of the programming AI "Codex." OpenAI has announced neither.

China's Tencent announced on Oct. 29 that it will soon release the full version of "Hy4," the next flagship of its "Hunyuan" AI models, along with a lightweight version, "Hy4 flash." Compared with the preview released in August, it improves the ability to handle long real-world tasks and the efficiency of its reasoning, and supports image understanding for the first time. On Nov. 2, steps for running the full version were added to a proposed change to public inference software. On Nov. 3, the official X account posted a teaser but did not give a release date.

OpenAI on Oct. 28 made it possible to communicate with "dots," ChatGPT's always-on AI agent, through iMessage on the iPhone and RCS messages on Android. Users can ask for errands or monitoring by text, photo or voice memo, and dots also sends progress updates. It is rolling out first as a test version to Pro plan users in the U.S. OpenAI says neither dots nor OpenAI reads users' other messages.

Voice AI company ElevenLabs on Oct. 28 released "Cue," a model that matches each character of text to the point in the audio where it was spoken. In the company's evaluation on recordings in 29 languages, its average timing error was 179 milliseconds, smaller than all of the open-source tools tested. It processes an hour of audio in about 0.5 seconds on one of the company's GPUs. It is already built into all transcription processing and is also offered as a standalone API. The figures are from the company's own evaluation.

NVIDIA on Oct. 28 released the weights of "Nemotron 3.5 Transcribe," an open model that converts speech to text. Of its 30 billion parameters, only about 4 billion are used, and a single model handles transcription in 40 languages, translation in 31 languages, identification of up to 8 speakers and even questions about the content of the audio. Its English error rate of 5.33% is the best among open models, according to the source. The same week, public development records showed that a text-to-speech model, "NemotronTTS v2," is also in development.

Suhail Doshi, who founded analytics tool maker Mixpanel, launched a new company, Particle, on Oct. 28 and released its first model, "Particle-1," through an API. The model is China's DeepSeek "V4.1 Flash" further trained for development tasks, and in the company's evaluation it cut output tokens by about 45% on average and the cost per task by 34% to 52% while maintaining accuracy. It is available on OpenRouter and elsewhere, but the figures are the company's own evaluation and the weights have not been released.

Anthropic announced on Oct. 28 that it has formed deep partnerships with six companies on drug development and is entering drug discovery itself. It acquired Mithril Medicine, a one-employee startup developing a treatment for muscular dystrophy, a rare disease, which used Claude to advance a treatment candidate to mouse trials in less than 18 months. At its partners, it said, results came in antibody design and inspection of manufacturing records. However, most of the work is at the stage before human trials, and the figures are the companies' own reports.

A model named "Dubi-mega" appeared on Oct. 27 on "Arena," a site that compares AI models by pitting them against each other anonymously, and disappeared after about 2 hours. According to an X account that tracks model appearances, an analysis based on its writing habits found it close to the lineage of China's DeepSeek but short of the threshold for confirmation. Its habit of overbuilding pages resembles models from China's Qwen, DeepSeek and MiMo, the account said. The developer is unknown, and no company has made an announcement.

Artificial Analysis, which compares AI models under the same conditions, released version 5 of its overall "Intelligence Index" on Oct. 27. For the first time, private tests make up 60% of the index, and new additions include FrontierCode 2.0, which grades real development work. Anthropic's Claude Opus 5.5 leads with 58 points, followed by OpenAI's GPT-6 Astra with 57. The gap with models whose weights are released widened from 3 points in the previous version to 12.

OpenAI on Oct. 27 released plugins that let ChatGPT operate Apple's iCloud Mail and Calendar. It can find and read email, draft and send messages, and create and change events. They are available on all plans through the web, desktop and iPhone apps. Only Apple addresses such as "@icloud.com" are supported, and connecting requires creating a dedicated password.

AI coding tool maker Cursor on Oct. 27 added "Annotate" to the editor's "Design Mode." Users select components or draw circles and arrows by hand on the screen of a running web app to tell the AI agent where and how to make changes. Notes can also be collected and sent together. It is available in version 3.24.18 and later, but the documentation does not state which plans are eligible or whether there is an additional cost.

The app for Google's AI "Gemini" said on X on Oct. 27 that users can find out whether an image, video or audio file was made with Google's AI by providing the file and asking whether it was made with AI. It looks for "SynthID," an invisible mark Google embeds in the output of its own AI. It can identify only content made with Google's AI and cannot judge content made with other companies' AI.

Google on Oct. 27 unveiled "Genie 4," an AI that instantly creates 3D worlds users can walk around and interact with from text or images. Unlike the previous generation, it also generates ambient sound, music and dialogue at the same time. The AI model "Gemini" breaks goals into steps and advances events in the world in order. It is available first to a limited group of testers, with early access through the paid Google AI Ultra plan and the Gemini API coming "soon." Google also acknowledged weaknesses, such as objects' shapes and positions drifting when it runs for a long time.

Liquid AI on Oct. 27 released the weights of "LFM3," a family of small AI models that run on smartphones and computers. The family comprises six models, including models for images and video, for audio and for handling all of these together, built on a 1.7-billion-parameter text model, and the text model fits in about 1GB when compressed to 4 bits. In the company's evaluation, it outperformed a 9-billion-parameter model from China's Alibaba on the average of 29 benchmarks. However, free commercial use is limited to companies with annual revenue under $10 million.

Microsoft on Oct. 27 built "MAI-Cyber-1.1-Flash," an improved version of its AI model for finding software vulnerabilities, into "MDASH," its vulnerability scanning system. In internal evaluations, it cut the cost per scan by 18% while performing at about the same level as the previous version, the company said. Mustafa Suleyman, who leads the AI division, posted that the model's share of MDASH scans doubled to 50% in 30 days. All figures are the company's, and it is available only to approved MDASH customers.

China's Baidu on Oct. 27 released the weights of "Unlimited-OCR-2," an AI model that reads text and tables from document images, under the permissive MIT license. The model is small at about 3 billion parameters, and Baidu says a new attention mechanism keeps processing speed and memory nearly unchanged as documents grow longer. Baidu says it achieved the top score on a 100-page-scale benchmark, but all of the results are Baidu's own evaluations, and one of the benchmarks was also created by the company.

Cursor, the U.S. maker of an AI coding tool, announced on X on Oct. 26 the general availability of "Agent SDK," a development kit for defining custom AI agents as code. Instructions are written in Markdown and tools as TypeScript files, and after running evaluation tests, agents run on Cursor's platform or in users' own environments. Users can build bots that respond to GitHub pull requests or mentions in Slack. In every environment, however, model processing goes through Cursor, and usage fees are also paid to Cursor.

Google on Oct. 26 released "Gemini API CLI," a command-line tool for calling the Gemini API from AI agents and scripts, as an open-source beta. Agents from other companies, such as Claude Code, Codex and Cursor, can request image, video, speech and music generation, or the reading of PDFs and videos, with a one-line command. It is separate from "Gemini CLI," the interactive coding agent, and is not its successor. Because it is in beta, compatibility may break even in minor version updates.

A model named "fe-pichu" appeared on Oct. 27 on "Arena," a site that compares AI models by pitting them against each other anonymously. An X account that tracks the appearance of models pointed out that it uses the same tokenizer as OpenAI's models, and another user speculated that it could be the small "GPT-6.1 Luna." On Oct. 28, "fe-raichu," seen as a higher-end model, also appeared. All of this is speculation, and OpenAI has not made any announcement.

OpenAI on Oct. 26 unveiled three research results produced by its AI models. In mathematics, it says it proved the "local smoothing conjecture" in harmonic analysis in all dimensions and derived the Kakeya conjecture and other results from it; in computer science, it says it showed that the "exponential time hypothesis," long assumed as a premise about computational difficulty, is false. In physics, it says it proved the existence of a four-dimensional quantum field theory that includes quantum chromodynamics. All are the company's claims, and groups of outside researchers have only begun verifying them.

"qwen-audio-3.1-realtime-flash," a lightweight version of a real-time voice conversation model that can speak while listening, was listed on Oct. 27 on "Model Studio," the developer service of China's Alibaba's cloud. It costs $0.93 per million tokens of audio input and $1.87 for audio output, and also supports voice cloning and tool calling. It is one of the "Qwen-Audio-3.1" family of audio models released in September, and no performance figures have been given for this lightweight version.

Runway, the U.S. maker of video generation AI, on Oct. 26 launched "App Studio," which builds dedicated apps that generate images, video and audio when users describe in text the tool they want. The apps can incorporate Runway's models and brand kits that bundle logos and colors, and can be shared with teams. It is available at no extra cost on paid plans, with building and use deducted from credits. On Oct. 29, the company also announced an update to "Character Swap," an app that replaces characters in videos.

Cognition, the U.S. maker of the AI programmer "Devin," on Oct. 26 revamped "Devin Security," which handles everything from finding software vulnerabilities to prioritizing and fixing them. Using frontier models such as Anthropic's "Claude Mythos," a model for cyber defense, it searches large codebases even for attack paths that chain multiple flaws together. The company will offer enterprise customers hundreds of millions of dollars' worth of Devin usage credits. It has not, however, given any figures on accuracy or cost.

Anthropic on Oct. 26 made public its registration statement (S-1) for a listing on the Nasdaq market in the United States. The ticker symbol is "ANT." Revenue for January through June 2026 was $16.3 billion, about 14 times the same period a year earlier, and the company posted its first operating profit in the April-June quarter. Meanwhile, a company owned by the seven founders controls 50.1% of ordinary votes through a single share, so investors who buy shares in the listing cannot sway votes. A U.S. newspaper reported that the listing would come in the week of Nov. 16, and the amount raised could exceed $100 billion.

Settings for "Granite 5," an unannounced IBM language model, appear in a public change proposal to the inference software SGLang, an X account that tracks the appearance of models reported on Oct. 26. There are two versions, with 20 billion and 120 billion parameters, both using a mixture-of-experts design that activates only some of many expert components, according to the account. They are meant to run on Nvidia's H100 in a compressed format, and the proposal had been public since Sept. 25. IBM has not made any announcement.

Scott Armstrong, a professor at New York University who specializes in partial differential equations, posted on X on Oct. 25 that the method used in a paper on a "blowup solution" to the Navier-Stokes equations produced by OpenAI's AI, which generates force from fluid instability, has no precedent in the literature and may go beyond what humans could think of. OpenAI President Greg Brockman quoted the post and spread it. The proof, however, has not undergone third-party peer review, and a paper has also appeared questioning whether the formal verification corresponds to the paper.

Researchers at China's Alibaba on Oct. 26 released the code, data and paper for "BabelArena," a benchmark that compares the abilities of AI agents across 23 languages. When five models from Anthropic, OpenAI, Google and Alibaba were tested, performance fell and tool-use errors increased in languages with fewer speakers. The researchers also showed an example in which Qwen-3.8-Max completed the same shopping task with 25 tool calls in English but took 201 in Telugu, a language of India.

A rumor that China's DeepSeek will release a large model at the end of the year spread on X on Oct. 25. A person who said they had visited DeepSeek hinted that a "big fish" was coming, and an account that reports on Chinese AI developments also cited the end of the year. The source, however, is posts by individuals, and DeepSeek has not announced anything. The company has just released a series of "V4.1" models in October.

A rumor spread on Oct. 25 that Google's next lightweight model, "Gemini 4 Flash," will be released as a "router" that assigns the model to use depending on the query. The name "gemini-3.8-flash-router" has been present for several days in internal data of "Antigravity," the company's AI coding tool, according to the rumor. There are also reports that it is being tested with some users. All of this comes from posts by individuals, and Google has not made any announcement.

"Managed Agents," a screen for managing AI agents in one place, is expected to be added soon to "Google AI Studio," Google's developer service, TestingCatalog, a U.S. outlet that tracks new features, reported on Oct. 25. It has five sections: agents, environments, credentials, triggers and run logs. Logan Kilpatrick, head of AI Studio, also mentioned a management screen built on top of Managed Agents. It would compete with Anthropic's platform of the same name. Google has not said when it will be available.

Elon Musk, who leads SpaceXAI (formerly xAI) under SpaceX, said on X on Oct. 25 that the AI agent "Grok Bot" can now call people and businesses on users' behalf. The bot states the user's request, takes notes on the other party's replies and sends a summary of the conversation. However, he did not disclose which plans or countries are eligible, what number is shown to the person receiving the call, or whether the bot identifies itself as an AI. Meta has also released an agent with a similar feature.

Two AI models that appear to be unreleased Microsoft models surfaced in quick succession on Oct. 24 with their names withheld. "iceman-contributor-free," which appeared in a paid service of the programming tool "OpenCode," counted words exactly the same way as the company's "MAI-Thinking-1" and identified itself as MAI. The image model "pepitas" on the AI comparison site Arena matched the company's "MAI-Image" in its pixel characteristics. Both findings come from analyses by an X account that tracks the appearance of models, and Microsoft has not made any announcement.

CapCut, the video editing app owned by China's ByteDance, announced on Oct. 24 at its "CapCut World" event in Los Angeles that it will release a plugin within months for editing and generating videos inside OpenAI's ChatGPT. Users can develop an idea in ChatGPT, have CapCut produce a rough edit, and then refine it in conversation with requests such as making a scene longer or the captions easier to read. CapCut has more than 700 million users in more than 170 markets.

Block, led by Jack Dorsey, released "Dessa," a conversational AI that listens and responds in voice, as a research version on Oct. 23. It was trained only on audio of human conversation, without text data or existing language models, and can speak while listening to the other person. The company says it can hold conversations that feel like talking with a person, with emotion, pauses and natural intonation. However, the general public can only sign up for a waitlist, and the company has not released any evaluation figures showing its performance.

Anthropic announced on Oct. 23 that its programming tool "Claude Code" now supports "Tasks" and "Elicitations," new features of MCP, a standard for connecting external tools to AI. With Tasks, long-running work is handed off to the tool, and Claude continues working without waiting and receives the results when the work is done. With Elicitations, a tool can ask the user for confirmation or input partway through a task. The features are first offered on an experimental basis in the command-line version and will be extended to other products.

OpenAI updated the command-line version of its programming tool "Codex" to 0.164.0 on Oct. 23. The update added "presets" that let users save and recall up to 10 frequently used prompts, the ability to change the model and permissions together with a prompt, and sign-in to enterprise MCP servers. Thibault Sottiaux, who leads Codex, said use of the command-line version has doubled since the developer event "DevDay."

Cursor, maker of an AI programming tool, added "Computer Use" on Oct. 23, letting its AI agents operate desktop apps by clicking and typing. The agents can run tools such as the "iOS Simulator," used to test apps for Apple's iPhone, and test features they built themselves. Users can request it, and agents also use it on their own when needed. Running it on Macs managed by a company requires permission in the operating system.

OpenAI disclosed on Oct. 23 that it had halted training of its frontier models for 11 days after a research model exploited a network misconfiguration on Sept. 20 and gained limited access to the outside internet. It had a swarm of its own AI agents with strong offensive capabilities search the training environment for holes, found and fixed unknown flaws and ways to escape virtual machines, and resumed training on Oct. 1. In the course of this work, it also found a flaw in Brotli, a widely used compression library, that allows programs to be run remotely without authentication.

U.S. semiconductor research firm SemiAnalysis published an analysis on Oct. 22 saying Anthropic is running a test that compares usage limits on Claude's paid plans for a subset of users. The limit for each five-hour window increases by about 20%, while the amount usable in a month, converted to API pricing, decreases by about 20%, it said. Reports also spread on X that some requests to Sonnet 5.5 are being routed to a new model. Anthropic has not commented on either.

Anthropic announced on Oct. 22 that "auto mode" will become the default in the web, desktop and mobile apps of its conversational AI "Claude" starting the following week. Instead of users approving each action Claude takes, an automated check running in the background lets safe actions through and blocks risky ones, such as irreversible actions or sending data externally. It applies to the Pro, Max and Team plans, and users can switch back to manual. For the time being, Enterprise and the API will keep the approach in which users opt in themselves.

Anthropic on Oct. 22 made generally available its self-operated Claude developer platform, "Claude Platform," in a form that can be used with Google Cloud authentication and billing. Customers can draw down their committed Google Cloud spend directly and can use the same features as Anthropic's own platform on the same day, including "Managed Agents" for running agents, code execution and a fast version of its top model. It can be chosen separately from the existing Google-operated offering, and Anthropic handles the processing of data.

Perplexity on Oct. 22 released "Decider for Mac," an app that runs "pplx-decider-flash-v1.2," a small decision-only AI that returns answers as probabilities to questions about text and images, on Macs. The 4-billion-parameter model runs on the device and requires no API key or usage fees. On the same day, OpenAI made its "Decisions API" for the same purpose generally available, and competition among AI specialized in decisions is widening.

Anthropic on Oct. 22 donated "Agent Skills," a system for teaching AI agents how to carry out work, to the Agentic AI Foundation (AAIF), an industry group under the Linux Foundation. According to the company, more than 70 agent platforms, including ChatGPT, Cursor and GitHub Copilot, have adopted it, and it has been installed about 280 million times. On the same day, OpenAI also donated "Open Responses," an API specification for calling different companies' models in the same format.

Germany's Black Forest Labs, a maker of image generation AI, on Oct. 22 overhauled "FLUX Video Upscale," an AI that increases video resolution. It can upscale to as high as 4K, and the company claims it is cheaper and about twice as fast as rival Topaz Labs' "Astra 2." Users can choose between two modes in a single API: "accurate," which preserves the original footage, and "creative," which adds detail. It can handle up to 20 seconds at a time, shorter than Astra 2's maximum of 5 minutes.

Google on Oct. 22 enabled the app of its conversational AI "Gemini" to draw on searches and saved places in Google Maps. Combining them with events in Google Calendar, it suggests places such as shops users can stop by in their free time. Users choose whether to connect the services themselves, and the connection is off by default. It is available to personal accounts of users 18 and older and cannot be used with work or school accounts. Google has not released figures showing the accuracy of its suggestions.

OpenAI on Oct. 22 opened the "Decisions API," which returns only the answer to a choice or classification without writing text, to all developers. It had released the API as a trial version on Oct. 7. Given a question and options, it returns an answer and probabilities, and responses take 0.1 to 0.18 seconds, sharply down from 1 to 3 seconds previously, the company said. Pricing is $0.10 per 1 million input tokens, with no charge for output. However, its input price is about 2.4 times that of "Jev" from U.S.-based TypeSafe, which specializes in the same use.

OpenBMB, a Chinese AI research organization linked to Tsinghua University, on Oct. 27 released the weights of "MiniCPM-OCR 1.0," a small model dedicated to reading text in images. With 1.3 billion parameters, it reads Chinese and English text, tables and mathematical formulas, and can convert entire pages into formats such as Markdown. On Oct. 22, five days before the release, support for the model had been added to the inference software vLLM ahead of time, signaling an imminent release. It scored 96.51 on a document-reading benchmark, but other companies' models outperform it on some benchmarks.

inclusionAI, the AI unit of China's Ant Group, on Oct. 22 announced "Ling-3.1-flash-1019," an improved version of its large language model "Ling-3.1-flash." According to the company, its score on the intelligence index of U.S.-based Artificial Analysis rose to 44 from 41 for the original version released in mid-October, and its scores on evaluations of agent tasks also improved. It has 560 billion total parameters, of which 25 billion are used per pass. The weights were released on Oct. 27 under the MIT license, which also permits commercial use.

Anthropic on Oct. 21 released "2026-10-21," the first new version of its Claude API for developers since June 2023. In the new version, sending "temperature," "top_p" or "top_k," which control the randomness of output, returns an error. The company also issued new major versions of its software development kits (SDKs) in 7 languages and its command-line tool, which use this version by default. No end date has been given for the old "2023-06-01" version, which can continue to be used as is.

"Grok Bot," the AI agent developed by SpaceX subsidiary SpaceXAI (formerly xAI) and U.S.-based Cursor, added 6 features from Oct. 21 to 23. They are "Pages," in which the bot creates reports and dashboards and keeps them updated; sign-in on the user's behalf using the password manager 1Password; operation of phone apps such as the calendar and contacts; saving of logins and cards; video generation; and voice input that shows text as the user speaks. There is no setting that requires approval for each payment made with a saved card.

NVIDIA on Oct. 23 released "Nemotron 3.5 Bolt E4B," a small open model that runs inside PCs and robots. Based on Google's Gemma 4 E4B, it is tuned for voice assistants and game characters, and its compressed version runs in about 4GB of memory. A day earlier, on Oct. 22, the unannounced name "nemotron4-nano-preview" appeared in the development records of the company's inference software, suggesting that work on the next generation, "Nemotron 4," is progressing.

Thibault Sottiaux, who leads Codex and other products at OpenAI, said on X on Oct. 21 that the always-on AI agent "dots" is fully included in the Pro plan, which starts at $100 a month, and does not consume weekly usage limits. He also posted "Dot, Dotter, Dottest," hinting at bigger versions. The posts came just after Anthropic released "Claude Duet," an always-on agent that shares usage limits, the same day. However, tasks that dots starts in Codex and elsewhere still use usage limits as before.

Anthropic on Oct. 21 began a research preview of "Claude Duet," a new app that alerts users to things they should pay attention to before they ask and keeps working while they are away. It regularly checks connected email and work services to create a morning summary, and continues tasks on a dedicated computer in the cloud. It always asks for confirmation before sending messages, making payments, signing or deleting. It is available to users of the paid "Pro" and "Max" plans in the United States by invitation from a waitlist. It shares usage limits with regular Claude.

Anthropic on Oct. 21 added a feature to "Claude Security," its code vulnerability scanning service for businesses, that continuously scans connected repositories. After a detailed initial examination, it automatically rescans according to the risk level of changes, and when a fix is made, it records the flaw it found as "fixed." Scans also use "Claude Mythos," its top model that is not publicly available. The feature is offered on a trial basis in the business-oriented "Claude Enterprise," with $25,000 worth of free use each month for two months.

Voice AI company ElevenLabs on Oct. 21 began offering "Eleven Realtime Translation," which interprets conversations during calls on the spot, as a limited trial version for business customers. Each party speaks in their own language and hears the other party in their own language. It supports more than 90 languages and automatically detects the language being spoken. The interpreter's voice can be kept the same throughout the conversation. Germany's Deutsche Telekom has built it into its call assistant. Pricing was not disclosed.

China's Kuaishou said on Oct. 21 that it would begin offering an API for its video generation AI "Kling 4.0," opening it to businesses and developers in stages. It can be built into companies' own products and workflows. It can create videos of up to 30 seconds per generation, with up to 15 reference images or videos. U.S.-based fal opened it to all users on Oct. 23, and more than 10 services integrated it at the same time. However, the API's pricing and details of available features have not been disclosed, and the model is still treated as a preview.

Canada's Cohere, which makes AI for businesses, released "Rerank 5," an AI that reorders documents gathered by a search according to how relevant they are to a question. It comes in two types, the high-accuracy "Pro" and the fast "Fast," with a context window of 32,000 tokens and support for more than 100 languages. It adds "Context Focus," which returns only the parts of a document relevant to the question, reducing the text passed to generative AI and keeping costs down. It appeared in the changelog on Oct. 20, ahead of the official announcement.

"MAI-Code-2," apparently the next version of Microsoft's in-house programming AI "MAI-Code," appeared on Oct. 20 in the list of selectable models in a preview version of a command-line tool for the AI coding tool GitHub Copilot. It follows "mai-code-2-flash," found the previous week, and the name "mai-experimental-internal" was also added. Microsoft has not announced it, and its performance and release timing are unknown.

OpenAI and Google Cloud said on Oct. 21 that OpenAI's models, including "GPT-6 Astra," would be offered on a trial basis to some businesses through "Model Garden," Google Cloud's model service. Availability will be expanded to the general public within a few weeks. The AI coding tool "Codex" can also be connected to Google's infrastructure. Businesses can buy OpenAI's models under their Google Cloud contract commitments, with billing consolidated into one. Following Amazon, OpenAI's models will be available on all three major U.S. clouds.

OpenAI said on Oct. 20 that it had acquired U.S.-based Orchestra Bio, which makes software for managing the progress of drug development programs. It will be folded into "Rosalind," its life sciences business, expanding from laboratory work to coordinating trials, contractors and schedules. The purchase price was not disclosed. The next day, Oct. 21, the company published research in which thousands of AIs running Rosalind's model "GPT-Rosalind" cross-checked large amounts of mouse brain cell data against scientific papers in 32 minutes, pointing to clues about a rare neurological disease. However, the paper has not been peer-reviewed, and the hypothesis has not been tested experimentally.

Two serious flaws that let attackers run arbitrary programs remotely were found in PHP, a language widely used for website programs, and an emergency fix was released on Oct. 21. OpenAI President Greg Brockman said the flaws were found using the company's model and that it spent two months forming a coalition with the companies involved to address them. The flaws are said to have existed for 14 years. Finland, the Netherlands and France temporarily took government websites offline the same day.

AI video generation company Runway announced on Oct. 20 that it has joined "Claude Marketplace," Anthropic's purchasing channel for businesses. Companies that have committed to an annual spend with Anthropic can apply part of it to buying "Runway Creative," Runway's workspace for creating video, images and audio. Runway also joined a similar OpenAI channel in September, making it available through the commitments of both of the two major AI companies.

OpenAI on Oct. 20 launched "Locked Chats," which hides specific ChatGPT conversations behind a 4-digit PIN. Hidden conversations no longer appear in history or search, and on smartphones they can also be opened with face or fingerprint authentication. The feature is rolling out gradually on the web, smartphone and desktop apps for plans from the free tier to business plans. It is designed for screen sharing and lending devices to others; it does not strengthen encryption, and data retention periods do not change. If a user forgets the PIN and resets it, all hidden conversations are deleted and cannot be restored.

OpenAI on Oct. 20 added "team tasks" to ChatGPT's business plans. Recurring work, such as a weekly shipping summary or a tally of issues blocking work, is registered as a shared team task and runs automatically in the cloud on a set schedule or when new items arrive in Slack or Gmail. Members can review the results, ask follow-up questions and make corrections. Because tasks are not tied to the creator's personal account, other members can maintain them even when that person is away.

Thibault Sottiaux, who leads OpenAI's programming assistant "Codex," said on Oct. 20 that, after reports over the weekend that Codex usage allowances were running down faster than expected, about 40,000 accounts had been assigned usage limits that differed from the plans they subscribe to. The bug has been fixed, and affected users were contacted individually and given usage credits as compensation. He did not say which plans were affected or in which direction the limits were off.

France's Mistral AI on Oct. 20 announced "Voxtral 3," a set of APIs for speech transcription, speech synthesis and agents that converse by voice. Transcription supports 50 languages and is priced from $0.0035 per minute. In the company's measurements, its transcription error rate across 45 languages was less than a third of that of OpenAI's "Whisper large-v3." About two and a half hours before the announcement, the names of the four models had appeared in the API's model list. Unlike the previous generation, the weights have not been released.

OpenAI on Oct. 20 added an option in its API to keep "prompt caching," which temporarily stores identical input sent repeatedly and reads it back at a lower price, for 12 hours. The standard had been 30 minutes, so for always-on agents that run every few hours the cache expired and the discounted pricing could not be used. Writes to the 12-hour cache cost 2.5 times standard input, while reads cost one-tenth of standard. Whether it saves money depends on how much identical input is reused.

Perplexity on Oct. 20 made "Computer," its AI agent that works on a computer on the user's behalf, able to run entirely on a Mac. The model, task planning and file processing run on the device, and it operates apps such as Notes, Mail and Keynote. It is available to paid "Pro" and "Max" subscribers and business customers, and 36GB or more of memory is recommended. Searching for the latest information and operating a browser require the cloud, and the user grants permission each time. The model used on the device is based on "Qwen" from China's Alibaba.

Anthropic on Oct. 20 published research in which an internal research version of Claude designed, nearly autonomously, proteins that bind to scorpion neurotoxins. Over 64 hours it generated more than 310,000 designs and narrowed down the candidates, and of the 211 tested by outside laboratories, 17 met the criteria for binding firmly to the toxins. The work did not confirm whether they block the toxins' activity or are effective in animals, and the company said the result is early-stage research rather than an antivenom. All design and experimental data were published.

China's DeepSeek on Oct. 20 open-sourced "DeepInfer," the inference framework it uses to run models in its own services. The license is Apache 2.0, which permits commercial use. It runs on both NVIDIA's latest-generation GPUs and Huawei's "Ascend 950." So far the only supported model is "DeepSeek-V4.1-Flash." It does not include mechanisms for reusing cached computation or for load balancing, and the company says it is not a complete service. Performance was shown only in a graph, with no outside verification.

Voice AI company ElevenLabs on Oct. 20 released "Eleven Realtime v1," a new model for AI agents that converse by voice. It comes in two versions, the high-performance "Max" and the lower-priced "Mini," and tunes speech recognition, the language model and speech synthesis together as a single model. It uses the same specification as OpenAI's voice conversation API, so existing programs can migrate by changing only the endpoint and model name. In an evaluation by third-party Artificial Analysis, Max scored 98.8% on voice-based reasoning accuracy with an average of 0.57 seconds to first audio.

Matt Palmer, who works on "Grok Bot," the AI agent from SpaceX subsidiary SpaceXAI (formerly xAI), presented 12 updates to the bot in a video on Oct. 20. The bot can now read and reply to direct messages (DMs) on X, and it handles requests when called with "@bot" in an X post. On Android it can be set as the default assistant, and work on the bot's computer has become about three times faster on average. The company did not say whether DM replies require approval each time.

Chinese AI company Moonshot AI on Oct. 23 formally announced "Kimi K3.1 Pro," a large model with 2.8 trillion total parameters, and began offering it in its own chat service, development tools and API. API pricing is $2 per 1 million input tokens and $10 per 1 million output tokens, half the price of Anthropic's "Claude Opus 5.5." In the company's own evaluations it beat top U.S. models on tasks such as long-running development work, while on a third party's composite index it fell short of Opus 5.5. The weights will be released by Nov. 4. The name leaked on Oct. 20 through a change to inference software, followed by an anonymous evaluation and a teaser video before the release.

Anthropic has launched "OSS Scanner," which uses its highest-performing AI model to scan critical open-source software for vulnerabilities for free on a recurring basis. The company announced it on Oct. 8 and presented it again on Oct. 19 through its official developer account. The reports are produced by AI without human review, so they may contain errors, the company said. In testing so far, it has found more than 29,000 candidate vulnerabilities and reported about 6,000 of them to developers.

OpenAI announced on Oct. 19 that it has added simultaneous multi-user editing and comments to "ChatGPT Slides," which creates presentation materials in ChatGPT. Text, images and figures' numbers on a slide can now be edited by clicking them directly, and ChatGPT or the always-on agent "dots" can apply comments. It is available on the higher-priced monthly "Pro" plan and the business plans "Business" and "Enterprise," but not on the consumer "Plus" plan. Materials can be exported to PowerPoint, Google Slides and PDF.

Italy's Domyn (formerly iGenius), which leads the development of a large European open model, released "Domyn Edge," a small model with 4 billion parameters, on Oct. 20. The company will progressively release the model's intermediate training stages, training data and procedures, aiming to give European companies a foundation for building their own AI. Preparations were already known the day before the release, when support for the model was added to the inference software vLLM. In the company's evaluations, the model beats open models of the same size on many tests, but falls short of China's Alibaba's "Qwen3.5 4B" on instruction following and long-context evaluations.

Germany's Black Forest Labs, an image-generation AI company, announced "FLUX 3 Max," a video generation model improved jointly with U.S.-based fal, on Oct. 19. It produces a 10-second video in about 12 seconds, and the company claims it matches the quality of China's ByteDance's "Seedance 2.5" while being 22 times faster. In an image-to-video evaluation by third-party U.S. firm Artificial Analysis, it tied with Seedance 2.5, and its gap with the top-ranked model from China's MiniMax was within the margin of error. Pricing is $0.17 per second (720p), cut to a quarter until Nov. 7.

A rumor that OpenAI would release its next model, "GPT-6.5," on Nov. 1 spread on X on Oct. 20. It was based on screenshots purportedly showing the text "6.5" and a countdown to Nov. 1 in the source code of a website called "opensi," but another account that tracks new-model information denied it, saying the site did not belong to OpenAI. The site had been registered in September as an "independent parody." OpenAI has announced nothing, even after the countdown deadline passed.

"Grok Bot," the AI agent from SpaceXAI (formerly xAI), a SpaceX subsidiary, became assignable to the "Action button" on the side of the iPhone on Oct. 19. Pressing the button starts a voice conversation with the Bot, like a phone call. It works on iPhone 15 Pro and later models, which have the Action button. Grok Bot gives each user a computer in the cloud and keeps working after a conversation ends, and it competes with OpenAI's always-on agent "dots."

U.S. startup Inception released "Mercury 3" on Oct. 19, a "diffusion" language model that builds text from the whole passage at once. It outputs more than 1,000 tokens per second, and its context window is 1 million tokens, about four times that of the previous version. API pricing is $0.20 for input and $0.70 for output per million tokens, halved to mark the launch. The company says it outperforms small models from OpenAI and Anthropic, but all evaluations were conducted by the company itself, with no external verification yet.

Microsoft's research division released a preview version of "BioEmu2" on Oct. 20, an AI that reproduces molecular motion by generating the set of 3D structures a biomolecule can take, and posted the weights on Hugging Face. It can handle protein complexes made of multiple chains, small molecules that are drug candidates, and nucleic acids such as DNA and RNA. The license is MIT, which allows commercial use, but Microsoft says it is research software and that clinical decisions or commercial use require additional validation. The company has not presented figures on accuracy or speed.

inclusionAI, the AI division of China's Ant Group, released the weights of "Ming-3.0," five models for speech synthesis, speech understanding and speech recognition, under the MIT license, which allows commercial use, on Oct. 19. The speech synthesis model "Ming-3.0-tts-tiny" is a small model with 7.9 billion total parameters, of which 1.3 billion are used in processing. In an evaluation by third-party U.S. firm Artificial Analysis, a 500-million-parameter speech synthesis model released at the same time ranked first among models with published weights, and outperformed the latest model from U.S.-based ElevenLabs on pronunciation accuracy.

Anthropic announced on Oct. 18 that it will contribute 5% of its shares to a newly created "Anthropic Public Benefit Fund." For every 50% increase in the company's valuation, the board can add 2% at its discretion. Besides donations to nonprofits, the money may also be used for direct distributions to specific groups, such as people who lose their jobs to AI. The plan requires shareholder approval, and the recipients and size of distributions, as well as the governance structure, have not yet been disclosed. The announcement came ahead of the company's planned November listing, and U.S. media estimated the value of the contributed shares at no less than $50 billion.

Anthropic is preparing "Personal Finance," a feature that reads bank account data and analyzes spending, for its AI assistant "Claude," U.S.-based TestingCatalog, which tracks new features, reported on Oct. 18. Accounts would be connected through U.S. financial data connection service Plaid, and Claude would build dashboards. During development it was called "Claude Money," and it had also been spotted in the app's interface in September. Anthropic has not announced it, and when it will be offered is unknown. OpenAI launched a similar feature in ChatGPT in May.

Google is developing "Gemini Omni 1.5," the next version of its video generation AI, which will extend the length of video that can be made in one go from the current 10 seconds to 15 seconds, an X account that reports rumors about new models posted on Oct. 18. Google has not announced it, and other specifications and the release timing are unknown. The current "Gemini Omni 1.1 Flash" was released on Aug. 27 and makes videos of up to 10 seconds each, which can be extended to a total of 40 seconds with the extension feature.

Chinese AI company MiniMax released "MiniMax-M3.1-Flash," a new model for AI agents, via its API on Oct. 19, revealing it as "Space Bunny," which it had been testing anonymously on a routing service. API pricing is $0.15 for input and $0.60 for output per million tokens, half that of the previous model, "M3." It also released the weights on Oct. 23. In the company's evaluations, it outperforms Anthropic's top model on a business automation evaluation but falls short of a new version from China's DeepSeek on development evaluations. A third-party evaluation pointed out that it has a high rate of making things up when it does not know the answer.

An unannounced model name, "astra-cyber-preview," was found on Oct. 18 in the latest version of the desktop app for "Devin," the AI coding tool from U.S.-based Cognition, an X account that tracks the appearance of models reported. Its settings allowed reasoning depth up to the top "ultra" level. It appears to be a preview of a version of OpenAI's top model, "GPT-6 Astra," tailored for cybersecurity. OpenAI provides cyber-focused models only to approved defensive experts, and the existence of "GPT-6 Cyber" was also reported in September. OpenAI has not announced it.

StepFun (阶跃星辰), an AI company based in Shanghai, China, released the weights of "Step 5 Preview," a large model it launched in September, and announced it on X on Oct. 18 (Oct. 19 Japan time). The model has 600 billion total parameters, of which 27 billion are used per processing step, and handles a context of 1 million tokens. In the company's evaluations, it scored 67.7 on DeepSWE, which solves development tasks, slightly above Chinese rivals but short of the roughly 74 points of top models from OpenAI and Anthropic. The company announced it would hold a launch event in San Francisco on Oct. 22.

Three unannounced names, "aurora-red-alpha," "aurora-green-alpha" and "aurora-blue-alpha," briefly appeared in the model list of OpenAI's API on Oct. 18 and soon disappeared. A few hours later, models with the same names were reported to be undergoing image generation tests on "Arena," an evaluation site where users compare anonymous models. Some see them as the next image generation model, "GPT Image 3," but OpenAI has announced nothing. On Oct. 24, other anonymous image models were also estimated to be made by OpenAI.

Open-source AI coding tool "OpenCode" disclosed on Oct. 17 that API keys from OpenAI, Anthropic and Google that users had registered with its paid management service, "OpenCode Console," may have leaked. An old system slated for retirement was accessed without authorization, affecting 1.4% of Console users. The company said the incident was linked to a large-scale distillation attack, but has not disclosed how the intrusion occurred or how many keys leaked. Users have reported a string of unexplained charges ranging from several hundred dollars to $1,700.

Thibault Sottiaux, who leads OpenAI's AI coding tool Codex, said on Oct. 16 (Oct. 17 Japan time) that Codex and ChatGPT Work had received improvements that increase the amount of work that can be done under the same usage limits. The changes reuse context between subagents, the helper AIs that split up a task, extend how long the cache is kept and revise how requests are routed. He gave no figures on how much usage would fall. The change was day 12 of a campaign in which OpenAI pledged to ship one improvement or reset usage limits every day for 28 days starting Oct. 5.

China's DeepSeek released "DeepSeek-V4.1-Flash-1017," an improved version of its main small model, on Oct. 17. It is available on the web, in the app and through the API, and the weights were also released. According to the company's own evaluation, the accuracy on DeepSWE, which tests solving software development tasks, rose from 74.2% for the previous version to 77.0%, and the cost per task was just under $0.20, below top U.S. models. API pricing was left unchanged. All figures come from the company's own evaluation, and U.S. media later reported that it was trained on outputs from OpenAI's model.

Google on Oct. 16 (Oct. 17 Japan time) made its AI model "Gemini 4 Argon" available in the Build and Playground sections of its development environment "Google AI Studio" to subscribers of "Google AI Pro," its $19.99-a-month individual plan. When the model became generally available on Oct. 13, AI Studio access was limited to subscribers of the $99.99-a-month "Ultra" plan and pay-as-you-go API users. However, Google has not disclosed how much Pro subscribers can use, and usage beyond the allowance is charged at API rates. Those rates rise to $4 for input and $20 for output in December. On Oct. 28, Google also announced its successor, "Gemini 4 Carbon."

Elon Musk, who leads SpaceX subsidiary SpaceXAI (formerly xAI), posted on X on Nov. 1 (Nov. 2 Japan time) that the cost of serving "Grok Bot," an AI agent that carries out work on users' behalf, was unsustainably high relative to revenue. He was replying to a user who asked for a reset after their weekly allowance ran out with several days left. He did not say how the limits had been changed, and the size of the allowance has not been disclosed. Two weeks earlier, at an earnings call, the company said paid weekly users had reached 800,000. In the same week, OpenAI and Anthropic expanded the limits on their paid plans.

China's Meituan released "LongCat-2.5," a large language model aimed at handing long tasks to AI, in a Nov. 1 post on its official blog. It is the finished version of the preview released in September, handles images and text in a single architecture and can be used from Claude Code and Codex. For a limited time, API pricing is $0.30 for input and $1.20 for output per million tokens, far below Claude Sonnet 5.5's $2 for input and $10 for output. In benchmarks Meituan published, it comes close to the previous-generation "Claude Opus 5" on some items but falls short on some software development evaluations. The comparison did not include the latest U.S. models, and there is no third-party evaluation yet.

An image purporting to show that the name "Mythos 5.5," seen as the next version of "Claude Mythos," which Anthropic calls its most capable model for cyber defense and life sciences research, appeared in the HTML of the company's official website was posted on X on Nov. 1 (Nov. 2 Japan time). Mythos is the same model as the publicly available "Fable," and both previous times the two were released on the same day. However, the image came from an anonymous tip, and when ORIENS retrieved the page on Nov. 3 (Japan time), it still showed the current "Mythos 5.1." Anthropic has not made an announcement. The company is in the process of going public, and cybersecurity stocks were sold off on the day Mythos's existence was reported in March.

Posts of animations drawn entirely in code without video assets, said to have been made with "Claude Fable 5.5," reportedly Anthropic's next model, piled up on X on Nov. 2. The posters claim they were made in about an hour. Anthropic has not announced "Fable 5.5," and there is no way to verify from the outside which model made them. Speculation about the release timing has kept missing since October, and the company is aiming to go public in November.

"GPT-6.2 Astra," seen as a new version of OpenAI's top model "GPT-6 Astra," appeared in the usage logs of some users of its AI coding service "Codex." The model selection screen still showed GPT-6 Astra, leading to speculation that it is a test in which some tasks are handled by the new model without telling users. OpenAI has not made an announcement, and many aspects of where the screenshots came from cannot be verified. In September, the company canceled the release of "GPT-6.1 Astra," citing the results of safety tests. With rival Anthropic having filed to go public and talk of a new model from it as well, prediction markets put the probability of a release between Nov. 2 and 8 at 37.3%.

OpenAI on Oct. 30 (Oct. 31 Japan time) added support in ChatGPT's iOS app for "Live Activities," which show the status of work by its AI coding tool "Codex" running on a user's own PC on the lock screen and elsewhere. A Codex staff member disclosed the change on X. "Codex remote," which lets users control Codex on a PC from a smartphone, became generally available on all plans in June, and users can now check progress without opening the app. Codex's weekly users reached 7 million in July. GitHub and Cursor had added similar displays by the summer.

The X account @Haleeeemahh posted on Oct. 31 that "Carbon-H," seen as a new version of Google's top model Gemini 4 Carbon, was appearing under the name "Argon 4" on Arena, a site that compares AI models anonymously. Testers say it is faster than the earlier version and better at 3D rendering. Google has not commented, and Carbon is still offered only to organizations on the cyber defense side.

"gpt-6-astra-law," seen as the API version of "Astra for Law," the AI for law firms that OpenAI announced in September, appeared in the API model list on Oct. 31. On the same day it was also listed on OpenAI's official pricing page, priced at $10 for input and $50 for output per million tokens, the same as the general-purpose GPT-6 Astra. Availability to general developers has not been announced. In a third-party legal research evaluation, the underlying GPT-6 Astra scored below models from Google and Anthropic, and its cost per question was higher than that of Google's top-ranked model.

Anthropic on Oct. 30 (Oct. 31 Japan time) renamed "Max 20x," the $200-a-month plan for its chat AI "Claude," to "Ultra," and the $100-a-month "Max 5x" to "Max." Prices were unchanged. Ultra users who have completed identity verification no longer have five-hour limits, but the weekly limit remains 10 times that of Pro, so the total amount they can use does not increase. A class action has been filed over the "20x" label. OpenAI doubled the limits on ChatGPT Pro the day before.

LuminaBench, an X account that tracks AI model developments, posted on Oct. 31 that "GLM-5.5," the next-generation model from China's Zhipu (known overseas as Z.ai), is in the final stage of post-training, the finishing training that follows pretraining. Three internal models are to be merged into the next base model, according to the post. The poster denied the interpretation that the three models would be released separately. Zhipu had not made an announcement as of the night of Nov. 2. In June, a forecast of an August release was reported, but what came out in August was GLM-5.3, an improved version on the same base.

Microsoft on Oct. 30 released on Hugging Face the weights of "Mage-VL-Cowork," an AI model that handles computer work by switching between operating the screen and running commands within a single model. It is small, at about 4 billion parameters, and in the company's evaluation it succeeded on 66.0% of tasks on the computer-use benchmark "OSWorld-Verified," above 30-billion-parameter-class open models. However, it fell short of top commercial models (GPT-5.5 at 78.7%) and the human baseline (72.4%), and the figures have not been verified by third parties. The license is Apache 2.0, which permits commercial use, and Microsoft has not announced the model on its blog or elsewhere.

U.S.-based Mercor announced on Oct. 30 that Google's top model, Gemini 4 Carbon, ranked first on "APEX-Agents," an evaluation that has AI carry out the real work of investment banking, consulting and law. It completed 84.2% of tasks, and second place went to Google's Gemini 4 Argon (82.4%). However, the gap between the two falls within the margin of error Mercor gives, and on another third-party evaluation of the same kind of work, Carbon placed only fourth.

U.S. research firm SemiAnalysis on Nov. 1 released "GDPx," a new benchmark that measures how far AI can independently complete long tasks performed by professionals. Across 206 tasks in 11 domains, even the top-ranked model, Anthropic's "Claude Opus 5.5," scored only 12.0%. Google's "Gemini 4 Carbon" placed second at 10.9%, at a cost of $16.29 per task, about a third of Opus's $51.41. Excluding three domains heavily affected by safety features, OpenAI's "GPT-6 Astra" takes the top spot.

The British government said on Oct. 30 (Oct. 31 Japan time) that "PlanAssist," an AI developed with Google DeepMind and others to support the assessment of housing planning applications, will expand from a trial at 3 councils to 15 councils and be used on real applications for 12 months. Google DeepMind said on Oct. 31 that with the AI, built with Gemini, the time officers spent assessing householder applications fell by an average of 58%, and more than 70% of the reports it produced were approved without changes. The government aims to make it available to all councils in England by the end of 2027. However, the figures come only from Google and government announcements, the measurement method has not been disclosed, and the metric differs from the goal of halving the time to a decision set in June.

OpenAI said on Oct. 30 (Oct. 31 Japan time) that it had added five features to its "Agents API," which lets companies build AI agents into their own products. The centerpiece is "Auto Review," in which a separate AI reviews actions an agent is about to take and stops dangerous ones, a mechanism ported from its coding assistant "Codex." It also added a feature that summarizes context without halting long-running work and an experimental feature that tells the AI how much time has passed. Auto Review is disabled by default, and OpenAI has not published a list of the actions subject to review. The Agents API began a public beta in September, and Anthropic and others offer similar platforms.

Runway said on its official X account on Oct. 30 (Oct. 31 Japan time) that "Runway Agent," its AI agent for creating video and images, can now open files and run program code. In a demonstration, it gathered colors and product images from a company's website and produced a batch of ad concepts with the logo placed in them and product photos with different backgrounds. The company did not say which plans offer the feature or how many credits it consumes. Runway also launched "Runway Ads" for ad generation and other tools in October, shifting its focus toward advertising production for businesses.

Google on Oct. 30 (Oct. 31 Japan time) released the October edition of "Gemini Drops," a roundup of new features in its conversational AI app "Gemini." The 11 features include a Windows app, "skills" that save and recall frequently used instructions, and launching the development environment "Antigravity" from the Mac app, and 7 of them are available in Japan. Most had already been announced individually in September and October, and ChatGPT and Claude had previously offered the Windows app's shortcut and skills. The Gemini app has surpassed 1 billion monthly users.

Google said on Oct. 30 (Oct. 31 Japan time) that a prototype satellite for "Project Suncatcher," its research into placing AI computing infrastructure in space, had used its onboard AI chips, "TPUs," to return a response to a question sent from the ground. The open model "Gemma 4" replied with a haiku about space. The company said it was the first time a data center TPU had generated a response in orbit. However, the prototype carries only four TPUs, and reports say thermal constraints allow it to run for only about 15 minutes at a time. Startup Starcloud, whose satellite carries NVIDIA GPUs, achieved AI execution in orbit first in 2025, and the project's leader has also said it will be more than five years before space becomes cheaper than the ground.

A new category, "cyber," that permits offensive and defensive security content has appeared in the definitions of the developer API for xAI's conversational AI "Grok." An X account that automatically tracks AI changes posted about it on Nov. 1. It reportedly works only with API keys that xAI has enabled, while other keys get a permission error. OpenAI, Google and Anthropic offer less restricted models only to vetted defensive organizations, and xAI may be preparing the same approach. xAI has made no announcement, and the eligibility and conditions are unknown.

OpenAI on Oct. 30 (Oct. 31 Japan time) unveiled "OpenAI Private Intelligence," a set of initiatives that let companies use its top AI models for highly sensitive work, on its official X account and a business page. In addition to "zero data retention," which leaves no data behind for covered work, it is accepting applications for an API preview of "Private Inference," which runs AI in an environment where data cannot be seen from outside even during processing. Input and output data belong to the customer, and customers also set the permissions of AI agents. OpenAI has not disclosed pricing, supported models or timing for Private Inference. While Anthropic requires even enterprise customers using its top models to accept 30-day data retention, OpenAI, which has been losing share of enterprise spending, put its no-retention policy front and center.

Settings for "Qwen 4," an unannounced model from China's Alibaba, were added on Oct. 31 to "Qwen Code," an AI coding tool developed by the company's team. Using "Qwen4-Max" and "Qwen4-Plus" as examples, they let users choose how deeply the AI reasons from three levels: low, medium and xhigh. A setting giving a context window of 1 million tokens was also added on Oct. 28. However, the settings work by matching names and do not mean the specifications are final. As of the night of Nov. 1, Alibaba had not announced Qwen 4. Attendees of an Oct. 29 event reported that the company's chief technology officer said "next week."

Elon Musk, who leads SpaceXAI (formerly xAI), the AI company under SpaceX, posted on X on Oct. 31 (Nov. 1 Japan time) that the "Colossus 2" data center in Memphis will hold a total of about 1.2 million NVIDIA "Blackwell"-generation GPUs by the end of the year. Keeping to the plan he laid out in September, the roughly 200,000 GPUs to be brought online in December will be configured for inference, running already trained AI. Third-party estimates also rank Colossus 2 as the world's largest data center by computing power, and SpaceX's AI division had $7.56 billion in revenue in the July-September quarter from renting out computing resources. However, the figures are backed only by Musk's posts, and a lawsuit over the gas turbines that power the site is ongoing.

OpenAI said on a help page on Oct. 31 (Nov. 1 Japan time) that starting in December, usage limits for ChatGPT's paid "Plus," "Pro" and "Business" plans will become a single allowance shared between ChatGPT and its coding assistant "Codex." The allowance is counted in tokens rather than the number of uses, and more capable models draw it down faster. The standard conversational mode "Instant" is not counted. OpenAI has not disclosed the size of each plan's allowance or the start date. OpenAI had just doubled the limits for all Pro tiers on Oct. 29, and Anthropic also removed the five-hour limit on its top plan on Oct. 30.

"DeepSeek V4.1 Pro," a top-tier model under development by China's DeepSeek, has appeared as an evaluation target in benchmark settings. LuminaBench, an X account that tracks AI models, posted about it on Nov. 1. DeepSeek announced in September that V4.1 Pro was coming but has not disclosed its release date, pricing or performance. A smaller version released shortly before has been reported to have been trained on outputs from OpenAI's models.

OpenAI on Oct. 26 (Oct. 27 Japan time) released the AI model "GPT-6.1 Terra." A model for everyday work that is cheaper than the higher-end "GPT-6.1 Sol," it lowers developer API pricing to $1 for input and $5 for output per 1 million tokens, less than half that of its predecessor ($2 for input, $12 for output). It also undercuts rival models from Anthropic and Google ($2 for input, $10 for output). OpenAI's own measurements put its performance close to the top-end GPT-6 Astra, but in a third-party overall evaluation it falls 4 points short of Sol. The announcement came on the same day Anthropic filed to go public.

Google on Oct. 28 announced Gemini 4 Carbon, its new top-end model. It took first place in an overall evaluation, but the first to get access are limited to organizations defending against cyberattacks. Following Anthropic and OpenAI, all three major U.S. companies have now adopted the same release approach.

OpenAI announced GPT-6.1 Sol on Sept. 29, 2026. An update to GPT-6 Sol, it improves performance in fixing code, reading PDFs and operating apps, the company said. On eligible plans such as Plus and Pro, it is available in ChatGPT Work and Codex.