Google launches Gemini 3.7 Flash for coding agents

Google DeepMind introduced Gemini 3.7 Flash, a new workhorse model focused on coding, agent workflows and knowledge-heavy tasks. Google says the model improves over Gemini 3.6 Flash in debugging, issue resolution, first-pass code accuracy, web development and business automation.

The most concrete signal is that Google is positioning 3.7 Flash as an agent model, not just a cheaper chat model. It is available through the Gemini API, Google AI Studio, Android Studio, Gemini Enterprise Agent Platform and Gemini Enterprise. Google also says Gemini Spark, its 24/7 personal agent inside the Gemini app for Pro and Ultra subscribers, is moving to 3.7 Flash.

Google claims the model is better at multi-step planning, tool use and recovering from roadblocks. In benchmarks cited by the company, 3.7 Flash beats 3.6 Flash on coding, web development, document reasoning and AutomationBench. Introductory pricing is 0.75 USD per million input tokens and 3.75 USD per million output tokens, half the original 3.6 Flash launch price per million tokens.

For developers and companies, the practical takeaway is simple: Google is trying to make its fast Gemini tier strong enough for production coding agents and workflow automation, where cost and reliability matter more than peak benchmark bragging rights.