Gemini 3.7 Flash arrives with a 1M token context and thinking modes

Google has released Gemini 3.7 Flash, the latest in its Gemini 3 series of natively multimodal reasoning models. It supports text, image, video, audio, and PDF inputs, with a 1,048,576-token input limit and a 65,536-token output limit. The model offers low, medium, and high thinking effort levels, but not minimal, and includes capabilities like code execution, computer use (preview), file search, function calling, and grounding with Google Search and Maps. It is available via the Batch API, Flex inference, and Priority inference, and a stable version is already out.