A router name in a development tool
A rumor 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 spread on X on Oct. 25 (the night of the same day, Japan time). The X account @LuminaBench reported it.
According to @LuminaBench, the name "gemini-3.8-flash-router" has been present for several days in internal data of Google's AI coding tool "Antigravity." There are also reports that it is already being tested with some users without notice (a gray test), according to the account. ORIENS has not been able to confirm these points, and Google has not made any announcement.
Current model is September's "Gemini 3.8 Flash"
The current lightweight model is "Gemini 3.8 Flash," released on Sept. 2. It has a context window of 1 million tokens and costs $0.75 per million tokens for input and $3.75 for output.
A router uses a light model for simple queries and a heavy model for difficult ones, holding down latency and cost. OpenAI also uses an approach in ChatGPT that automatically routes queries to models depending on the query. Because the name includes "3.8," it may be routing centered on the current Flash, and its relationship to Gemini 4 Flash is unclear.
Gemini 4 rolls out from the top down
Google officially announced the higher-end "Gemini 4 Carbon" on Oct. 29. "Gemini 4 Argon" had also been released before that, and some expect the lightweight Flash to follow.
