Introducing Gemini 3.7 Flash

Google has released Gemini 3.7 Flash, positioned as its most intelligent workhorse model designed specifically for coding and agents. Arriving just three weeks after the previous version, the new model introduces major performance improvements across software engineering, web development, and knowledge-dense fields like law, finance, and biosciences.

The model achieves higher accuracy in software debugging, issue resolution, and generating production-ready code. It also creates more functional layouts and feature-complete applications in web development using fewer prompts. In complex knowledge tasks, it shows better reasoning and document-processing capabilities, alongside improved performance in real-world business workflows. Additionally, the model provides an enhanced developer experience by adapting better to roadblocks, engaging in multi-step planning, and following instructions with higher fidelity.

Through the end of the year, Gemini 3.7 Flash is available at an introductory price of half the original 3.6 Flash cost, charging $0.75 per million input tokens and $3.75 per million output tokens. This reduced pricing is intended to help developers and customers scale production-ready agents cost-effectively.

The model is also being integrated into Gemini Spark, the personal AI agent available to Google AI Pro and Ultra subscribers, to improve efficiency in knowledge work and Google Workspace tool use. Furthermore, Gemini 3.7 Flash ships with updated safety safeguards designed to prevent misuse in chemical, biological, radiological, nuclear, and cyber offense domains while supporting beneficial applications.