MOUNTAIN VIEW, Calif.: Google launches Gemini 3.7 Flash, an AI model built for software coding and automated business workflows, at half its predecessor’s price.
The Thursday release arrives three weeks after Gemini 3.6 Flash. Google attributes the unusually short turnaround to developer feedback and algorithmic improvements. The company calls it “our most intelligent workhorse model yet” for coding and agents.
Introductory pricing runs $0.75 per million input tokens and $3.75 for output through the end of 2026. Rates double to $1.50 and $7.50 on January 1, 2027. The current price is half the original cost of Gemini 3.6 Flash.
Google reports significant benchmark gains over the prior model. Gemini 3.7 Flash scores 43.6% on FrontierCode 1.1 Main, up from 34.4%. On DeepSWE v1.1 it reaches 65.3%, versus 49% previously. On AutomationBench, which measures enterprise workflow automation, Google’s table shows 30.4% against 23.6% for OpenAI’s GPT-5.6 Terra and 10.7% for Anthropic’s Claude Sonnet 5. The figures are Google’s own measurements.
The model is designed to plan tasks, use software tools, and complete multi-step workflows with less human intervention. Google says it improves debugging, issue resolution, and first-pass code generation, and reproduces web designs from screenshots with fewer iterations.
Availability spans the Gemini API, Google Antigravity, AI Studio, Android Studio, and GitHub Copilot. Gemini Spark, Google’s subscription agent service in more than 160 countries, runs on the new model starting Thursday.
Google offered no timing for Gemini 3.5 Pro, its awaited flagship. The launch follows last week’s leadership overhaul at Google DeepMind and a Reuters report that co-founder Sergey Brin has urged key AI staff to focus fully on Gemini.
Previous
