Google DeepMind releases Gemini 3.7 Flash with improved coding, agentic, and document-processing capabilities

Google DeepMind today introduced Gemini 3.7 Flash, positioning it as the company’s most intelligent workhorse model for coding and agentic workflows. The release follows the prior 3.7 Flash iteration by three weeks and incorporates developer feedback alongside algorithmic improvements slated for future models.

In coding benchmarks, 3.7 Flash outperforms its predecessor on FrontierCode 1.1 Main (43.6% vs 34.4%) and DeepSWE v1.1 (65.3% vs 49.0%), indicating stronger debugging, issue resolution, and production-ready code generation. For web development, the model shows higher design adherence and parity when generating layouts and feature-complete apps from reference inputs such as screenshots or design systems, and it surpasses 3.6 Flash on Arena.ai’s WebDev Arena with an Elo score of 1588 vs 1538.

In knowledge-dense domains, 3.7 Flash improves reasoning and accuracy on the GDP.pdf benchmark (34.0% vs 22.0%) and AutomationBench (30.4% vs 17.0%), reflecting better handling of complex documents and real-world business workflows.

Gemini 3.7 Flash is available through the end of the year at an introductory price of $0.75 per million input tokens and $3.75 per million output tokens, which is half the cost of the original 3.6 Flash pricing. Google cited early customer feedback highlighting the model’s performance and precision at reduced cost.

Google also announced that Gemini Spark, its 24/7 personal AI agent available to Google AI Pro and Ultra subscribers in over 160 countries, will begin using 3.7 Flash today. The update aims to improve tool use for Google Workspace apps and the quality of outputs for complex, multi-skill workflows such as consolidating files, drafting emails, and updating status documents.

The company emphasized safety improvements in Gemini 3.7 Flash, including updated safeguards against misuse in Chemical, Biological, Radiological, and Nuclear (CBRN) and cyber offense domains, while enabling beneficial use cases consistent with its bioresilience and cyber programs.