Google launches Gemini 3.7 Flash for coding and agent workflows

Google has launched Gemini 3.7 Flash, a new workhorse model aimed at coding, agent workflows, web development and document-heavy professional tasks.

A rapid follow-up to Gemini 3.6 Flash

The release arrives only three weeks after Gemini 3.6 Flash. Google says the newer model incorporates developer feedback and algorithmic changes intended to improve software engineering, knowledge work and multi-step execution.

Stronger coding performance

Google reports that Gemini 3.7 Flash improves debugging, issue resolution and first-pass code accuracy. On FrontierCode 1.1 Main, the company lists a score of 43.6 percent compared with 34.4 percent for Gemini 3.6 Flash. On DeepSWE v1.1, Google reports 65.3 percent versus 49.0 percent.

These benchmark figures are provided by Google and the benchmark operators and may not reflect every production codebase or workflow.

Better web development and UI generation

Google says the model can produce more complete web applications in fewer prompts and follow visual references more closely. On WebDev Arena, Google reports an Elo score of 1588 for 3.7 Flash compared with 1538 for 3.6 Flash.

More capable agent workflows

The model is designed to plan across multiple steps, recover from roadblocks and use tools with less manual supervision. Google also highlights improvements in enterprise workflow automation and complex document processing.

Introductory pricing

Gemini 3.7 Flash is available through the end of 2026 at an introductory price of $0.75 per million input tokens and $3.75 per million output tokens. Google says pricing will increase to $1.50 per million input tokens and $7.50 per million output tokens starting January 1, 2027.

Where developers can use it

The model is available through the Gemini API in Google AI Studio and Android Studio, as well as Google Antigravity and Gemini Enterprise. Google has also moved Gemini Spark, its always-on personal agent for eligible subscribers, to Gemini 3.7 Flash.

Safety updates

Google says the release includes updated safeguards for cyber offense and chemical, biological, radiological and nuclear misuse. The company has published a model card with additional information about its evaluations and safety approach.

Practical significance

Gemini 3.7 Flash shows how quickly the competition is shifting toward models that balance intelligence, latency and cost. For developers building production agents, a smaller improvement in task completion can matter substantially when it reduces retries, tool calls and total token usage across large workloads.

Google announced the model in an official product post .

Source: Google