Google has released Gemini 3.7 Flash, an updated model designed to enhance coding, agent workflows, and knowledge work capabilities. This release follows the previous version by three weeks and incorporates developer feedback alongside algorithmic innovations.
- Achieves higher first-pass code accuracy, scoring 43.6% on FrontierCode 1.1 Main and 65.3% on DeepSWE v1.1 compared to 3.6 Flash.
- Outperforms the previous model on Arena.ai’s WebDev Arena with an Elo score of 1588 versus 1538.
- Demonstrates improved reasoning in complex documents, reaching 34.0% on the GDP.pdf benchmark and 30.4% on AutomationBench.
- Available at an introductory price of $0.75/1M input tokens and $3.75/1M output tokens through the end of the year.
- Powers Gemini Spark for Google AI Pro and Ultra subscribers, improving tool use for Google Workspace apps.
The update enables developers to scale production-ready agents more cost-effectively while providing enhanced safety safeguards against CBRN and cyber offense misuse.