A free model with an undisclosed developer
"Pulse Alpha," a "stealth" model whose developer is not disclosed, became available for free on the AI coding tool OpenCode on Nov. 2. @aitrackerbot, an X account that tracks new model releases, reported it the same day (Nov. 3 Japan time). The model reportedly supports reasoning, image input and tool calling.
According to OpenCode's public data, it has a context window of 1 million tokens and a maximum output of 131,000 tokens, and accepts text and images. Usage rose from 6.1 billion tokens on Nov. 2 to 109 billion tokens on Nov. 3, with about 2,500 users.
Guesses: Tencent, Qwen, GLM
Users were divided over its identity. @LuminaBench, which runs benchmarks, named Tencent's "Hy4" or some version of Alibaba's Qwen as its initial guess. Others suggested it was GLM 5.5 from China's Zhipu.
Analysis points to a new Thinking Machines model
About an hour later, @aitrackerbot followed up with an analysis saying it was likely a new model from Thinking Machines Lab. It said it compared how tool definitions are written against 11 other labs and about 500 chat formats, and that the billing method of counting images in 40-by-40-pixel units matched the company's model "Inkling." It said, however, that the model is not Inkling itself.
Thinking Machines Lab is a U.S. startup led by Mira Murati, former chief technology officer of OpenAI. On July 15 it released Inkling, its first model with publicly released weights. It is a large model with 975 billion total parameters, 41 billion of which are used at a time, and accepts text, images and audio.
The developer of Pulse Alpha has not confirmed any of these guesses. In August, OpenCode also offered a mystery model, "Ox Alpha," for free, which drew attention.
