On September 3, 2026, Google announced the release of its new model, "Gemini 3.8 Flash." This model was introduced just a few weeks after the previous generation, Gemini 3.7 Flash.
Model ReleasesGoogleGemini 3.8 Flash
Google Releases Gemini 3.8 Flash: Increased Reasoning Steps for Complex Tasks and Iterative Tool Calling
This article is a translation. Read the Japanese original
The company explained that the new model executes more reasoning steps for complex tasks. It also stated that the model possesses the capability to call tools iteratively.
Pricing remains the same as the previous generation, with the initial pricing structure maintained: $0.75 per 1 million input tokens and $3.75 per 1 million output tokens.
Source: Google says its new Gemini 3.8 Flash model ‘works harder’ but might cost more (The Verge AI, 2026-09-03)