Google's Gemini 3.7 Flash Leads on Coding Benchmarks, Priced at Half Its Predecessor

Google launched Gemini 3.7 Flash on August 13, positioning it as the company’s primary workhorse model for coding and agent workflows. The release comes three weeks after Gemini 3.6 Flash and lands at introductory pricing of $0.75 per million input tokens, roughly half what 3.6 Flash cost.

On FrontierCode 1.1 Main, which evaluates code across multiple languages with requirements for bug testing and style compliance, Gemini 3.7 Flash scored 43.6%, up from 34.4% on the previous version. On DeepSWE v1.1, it reached 65.3%, compared to 49.0% for 3.6 Flash. Google also cited a 34% score on the GDP.pdf business document benchmark, putting it 6% ahead of Claude Sonnet 5 and 9.3% ahead of GPT-5.6 Terra on that test.

The model supports a one million token context window and generates responses up to 64,000 tokens. Google says it’s particularly strong at UI generation, with better alignment to reference images than its predecessor, and supports ensemble agent applications with improved multi-step planning and instruction following.

The architecture is the same as 3.6 Flash: a transformer-based mixture-of-experts design derived from Gemini 3 Pro. No changes to training methodology are disclosed in the model card.

Gemini Spark, Google’s 24/7 personal agent for Pro and Ultra subscribers in over 160 countries, runs on 3.7 Flash from launch day. Google says Spark becomes more efficient at knowledge work with the upgrade, including better tool use across Google Workspace apps for tasks like consolidating files, drafting emails, and updating status documents.

Gemini 3.7 Flash is available in Google AI Studio, Android Studio, and the Antigravity platform.

Pricing: $0.75 per million input tokens and $3.75 per million output tokens through December 31, 2026. Starting January 1, 2027, those rates double to $1.50 and $7.50 respectively.

Sources: