Google Ships Gemini 3.7 Flash, a Half-Price Workhorse for Coding and Agents

Three weeks after Gemini 3.6 Flash, Google's newest workhorse model posts steep gains in coding and agent benchmarks while cutting the introductory token price in half.

Three weeks after shipping Gemini 3.6 Flash, Google is back with Gemini 3.7 Flash — what it calls its "most intelligent workhorse model yet for coding and agents." Announced August 13, the release arrives at half the previous model's introductory price while posting the series' biggest benchmark jumps across software engineering, web development, and knowledge work.

On coding, 3.7 Flash lifts Google's DeepSWE v1.1 score from 49.0% to 65.3% and FrontierCode 1.1 Main from 34.4% to 43.6%, Google said in its launch post. Web development climbs too — the model's WebDev Arena Elo rises from 1538 to 1588, and Google reports it "generates more functional layouts and feature-complete apps in fewer prompts." In knowledge-dense fields like finance, law, and biosciences, its document-processing benchmark score jumps from 22.0% to 34.0% while its business-workflow score nearly doubles from 17.0% to 30.4%, 9to5Google reported.

The economics are the sharper signal. Through the end of 2026, Gemini 3.7 Flash costs $0.75 per million input tokens and $3.75 per million output tokens — half the launch price of 3.6 Flash. It rolls out starting today in Gemini Spark for AI Pro and Ultra subscribers, Google Antigravity, AI Studio, Android Studio, and the Gemini Enterprise Agent Platform, and it ships with updated safeguards against chemical, biological, radiological, nuclear, and cyber-offense misuse.

Sources: Google, 9to5Google