Overview
Gemini 3.7 Flash is a natively multimodal reasoning model in the Gemini 3 family, positioned as a workhorse for coding and agentic workflows. It is based on Gemini 3.6 Flash and was announced on August 13, 2026. (blog.google)
Capabilities and uses
- Accepts text, images, video, audio and PDFs; produces text. The documented input limit is 1,048,576 tokens and the output limit is 65,536 tokens.
- Supports configurable low, medium and high thinking levels, function calling, code execution, structured output, context caching, file search, URL context, computer use, and grounding with Google Search and Maps.
- Intended uses include software engineering, debugging, web development, document analysis and multi-step agent workflows. (aistudio.google.com)
Limitations and availability
The model does not support image generation, audio generation or the Live API. The minimal thinking level is unsupported and returns an error. Its knowledge cutoff is March 2026, although knowledge in some domains may be limited to January 2025. Outputs should be checked for important decisions. (aistudio.google.com)
As checked on November 16, 2026, Google’s Gemini API documentation says the model is deprecated, is no longer selectable in Google AI Studio, and API requests to gemini-3.7-flash are automatically routed to Gemini 3.8 Flash. Separately, Google Cloud’s Gemini Enterprise Agent Platform documentation schedules retirement of Gemini 3.7 Flash for January 28, 2027. These are platform-specific availability conditions. (aistudio.google.com)
Pricing
Google’s launch announcement listed an introductory price in USD of $0.75 per 1 million input tokens and $3.75 per 1 million output tokens, expiring December 31, 2026. It announced subsequent rates of $1.50 per 1 million input tokens and $7.50 per 1 million output tokens beginning January 1, 2027. These are the announced rates for Gemini 3.7 Flash, not a verified quote for requests now routed to Gemini 3.8 Flash; check the applicable platform’s current billing terms. Pricing source checked November 16, 2026. (blog.google)