Gemini 3.7 Flash: workhorse for coding and agents, intro price cut in half
Google ships Gemini 3.7 Flash on 13 Aug 2026. Clear coding and agent gains: FrontierCode 43.6%, DeepSWE 65.3%. Introductory price $0.75/$3.75 per 1M tokens through the end of 2026.
Google announced Gemini 3.7 Flash on 13 Aug 2026, a Flash-tier model aimed at coding and agents. Benchmark scores rise over 3.6 Flash, and through the end of 2026 the price is half of the standard rate that starts in 2027.
Where do coding and agents improve?
On FrontierCode 1.1 Main, 3.7 Flash scores 43.6%, up from 34.4% on 3.6 Flash. DeepSWE v1.1 rises from 49.0% to 65.3%. WebDev Arena Elo is 1588 versus 1538, GDP.pdf reaches 34.0% versus 22.0%, and AutomationBench reaches 30.4% versus 17.0%.
The API model code is gemini-3.7-flash. Token limits are 1,048,576 input and 65,536 output. The model supports thinking at three levels (low, medium, high; no minimal), function calling, computer use (preview) and search grounding.
Intro pricing and how long does it last?
Through 31 Dec 2026, the introductory price is $0.75 per 1M input tokens and $3.75 per 1M output tokens. From 1 Jan 2027 it doubles to $1.50 and $7.50.
The model is available in the Gemini API (via Google AI Studio and Android Studio), Google Antigravity, the Gemini Enterprise Agent Platform and the Gemini Enterprise app. In the Gemini app it powers Spark, the 24/7 personal agent for Google AI Pro and Ultra subscribers.