Google’s newest Flash model beats 3.6 Flash on coding, documents, and workflows — and costs half as much through the end of 2026.
Google introduces Gemini 3.7 Flash as its most intelligent workhorse model yet for coding and agents, arriving just three weeks after Gemini 3.6 Flash. The August 13, 2026 announcement from Tulsee Doshi and the Gemini team is presented as a response to developer feedback plus new algorithmic work Google plans to carry into later models. Gemini 3.7 Flash targets software engineering, knowledge work, and web development, and it launches at an introductory price of half the original 3.6 Flash cost per million tokens.
The model posts clear gains over 3.6 Flash. FrontierCode 1.1 Main production-code quality rises from 34.4% to 43.6%, and DeepSWE v1.1 long-horizon software engineering jumps from 49.0% to 65.3%. WebDev Arena Elo moves from 1538 to 1588, with more functional layouts and feature-complete apps in fewer prompts. On complex documents, GDP.pdf improves from 22.0% to 34.0%. AutomationBench, which tests real business workflows, goes from 17.0% to 30.4%. Google also says first-pass code accuracy, debugging, instruction following, and multi-step tool use are stronger, so teams should need fewer retries.
Through December 31, 2026, Gemini 3.7 Flash costs $0.75 per million input tokens and $3.75 per million output tokens; from January 1, 2027 the rates become $1.50 and $7.50. Developers can use it in Google AI Studio, the Gemini API, Android Studio, and Google Antigravity. Enterprises get it in Gemini Enterprise and the Agent Platform. Google AI Pro and Ultra subscribers in more than 160 countries receive the update through Gemini Spark, the 24/7 personal agent, with better Workspace tool use for email, files, and status documents. The release includes updated Frontier Safety safeguards against CBRN and cyber-offense misuse. Read more on the Google Keyword blog.
Original Sources
Disclaimer
This is an AI-summarized article from authoritative sources. If you want your content removed, please .