Introducing Gemini 3.7 Flash

Released August 13, 2026, Gemini 3.7 Flash arrives just three weeks after 3.6 Flash, built on developer feedback and algorithmic innovations. It is positioned as Google's most intelligent workhorse model for coding and agents, with substantial improvements across software engineering, knowledge work, and web development workflows.

In coding, 3.7 Flash improves debugging and issue resolution, with higher first-pass code accuracy and more production-ready output, scoring 43.6% vs 34.4% on FrontierCode 1.1 Main and 65.3% vs 49.0% on DeepSWE v1.1. For web development, it generates more functional layouts and feature-complete apps in fewer prompts, achieving an Elo of 1588 vs 1538 on WebDev Arena, with high design adherence whether the reference is a screenshot, image, or full design system.

For knowledge-dense fields, 3.7 Flash significantly outperforms 3.6 Flash on GDP.pdf (34.0% vs 22.0%) and AutomationBench (30.4% vs 17.0%), showing better reasoning in finance, law, and biosciences and more effective completion of real-world business workflows. It also adapts better to roadblocks, clarifies intent, and follows instructions with greater fidelity, putting more effort into multi-step planning and tool calls to reduce manual oversight and retries.

The model is available at an introductory price of $0.75/1M input tokens and $3.75/1M output tokens through the end of the year, roughly half the original 3.6 Flash cost. Early customer feedback highlights its performance and precision at low cost, and Gemini 3.7 Flash now powers Gemini Spark for Google AI Pro and Ultra subscribers in over 160 countries, a 24/7 personal AI agent.