Gemini 3.7 Flash release notes / what's new

Google DeepMind has introduced Gemini 3.7 Flash, a high-intelligence "workhorse" model specifically optimized for coding and agentic workflows. Released on August 13, 2026, this model provides substantial improvements in software engineering, web development, and knowledge work while reducing the introductory cost per million tokens by half compared to Gemini 3.6 Flash.

Enhanced Performance in Coding and Software Engineering

Gemini 3.7 Flash demonstrates significant gains in debugging, issue resolution, and the generation of production-ready code. It achieves higher first-pass accuracy than its predecessor, as evidenced by the following benchmarks:

  • FrontierCode 1.1 Main: 43.6% (up from 34.4% for 3.6 Flash)
  • DeepSWE v1.1: 65.3% (up from 49.0% for 3.6 Flash)

Web Development and UI Generation

In web development workflows, Gemini 3.7 Flash generates feature-complete applications and functional layouts using fewer prompts. The model exhibits high design adherence when generating UIs based on reference inputs such as screenshots, images, or full design systems. On the Arena.ai WebDev Arena, Gemini 3.7 Flash achieved an Elo score of 1588, compared to 1538 for Gemini 3.6 Flash.

Reasoning in Knowledge-Dense Fields

Gemini 3.7 Flash improves reasoning and accuracy for complex documents in specialized fields such as law, finance, and biosciences. Key performance improvements include:

  • GDP.pdf benchmark: 34.0% accuracy (compared to 22.0% for 3.6 Flash), testing the model's ability to process complex documents.
  • AutomationBench: 30.4% success rate (compared to 17.0% for 3.6 Flash), demonstrating improved capability in completing real-world business workflows.

Developer Experience and Pricing

Gemini 3.7 Flash is designed for more disciplined execution, featuring improved multi-step planning, tool calling, and instruction following. This reduces the need for manual oversight and retries in engineering workflows.

Through the end of the year, the model is available at an introductory price of:

  • Input tokens: $0.75 per 1 million tokens
  • Output tokens: $3.75 per 1 million tokens

Integration with Gemini Spark

Gemini 3.7 Flash is now the underlying model for Gemini Spark, the 24/7 personal AI agent available to Google AI Pro and Ultra subscribers. This update enhances Spark's efficiency in knowledge work, specifically improving tool use for Google Workspace apps to better consolidate files, draft emails, and update status documents.

Safety and Safeguards

Gemini 3.7 Flash includes updated Frontier Safety safeguards to prevent misuse in cyber offense and Chemical, Biological, Radiological, and Nuclear (CBRN) domains, while continuing to support beneficial use cases in alignment with Google's cyber program and bioresilience approach.

Availability

  • Developers: Available via Google AI Studio, Android Studio, and Google Antigravity.
  • Enterprises: Accessible through the Gemini Enterprise app and the Gemini Enterprise Agent Platform.
  • Individuals: Available via Gemini Spark in the Gemini app for subscribers in supported countries.

Sources

Related