Google Releases Gemini 3.7 Flash, Its Most Advanced Flash Model to Date

The latest Flash Model is designed for coding and agentic workflows, with stronger performance and lower costs at scale.

Shortly after releasing three updated Gemini Flash Models, Google has already taken another step forward in making agentic AI more capable and affordable.

Gemini 3.7 Flash is built for coding, UI design, and complex knowledge work, with stronger agentic capabilities that help AI take on more of a workflow from start to finish. Google says its “most intelligent workhorse model yet” builds on real-world developer use and technical upgrades that improve how it handles advanced tasks.

For businesses, that combination of speed, intelligence, and efficiency brings agentic AI one step closer to becoming business as usual.

Details of the Update

Gemini 3.7 Flash continues where the most recent updates left off, once again making significant improvements to workflow efficiency.

The model is designed to perform better across longer, more demanding tasks, especially those involving building interfaces, coding, and working through information-heavy material.

The most noticeable improvements include:

• Coding: Once again, Gemini Flash improved at writing, debugging, and refining code. Google highlighted improved first-attempt accuracy for coding and stronger results on software engineering benchmarks, including a jump from 49% to 65.3% on DeepSWE v1.1.
• Business automation and agentic workflows: Gemini 3.7 Flash scored 30.4% on AutomationBench, up from 17% for Gemini 3.6 Flash. The model showed stronger autonomous performance across multi-step business processes, changing its approach when conditions or tasks changed direction or additional input was needed.
• UI and web creation: The model demonstrates stronger performance when developing interfaces and designing apps, with stronger visual-to-interface conversion and fewer prompts required to finish tasks.
• Complex analysis: Google reported gains in reasoning and working through difficult documents across specialized fields, such as law, biosciences, and finance.
• Pricing: Gemini 3.7 Flash will cost just $0.75 per million input tokens and $3.75 per million output tokens through December 31, 2026, which is half the original price of Gemini 3.6 Flash.

Together, these changes suggest Google is thinking far beyond speed. Gemini 3.7 Flash demonstrates the ability to stay on task across multiple steps, adapt as requirements change, and manage more of the workflow with less human input.

Gemini 3.7 Flash brings agentic AI another step closer to handling everyday work at scale.

From an AI Advantage to a Workflow Essential

Google is moving fast, releasing tools designed to take on more of the workflows businesses rely on every day. Brands that want a competitive advantage will need to adapt just as quickly.

Agentic AI raises the stakes because companies can use emerging technology for more than just getting answers. Businesses and teams can use these tools to plan, test, and carry out multi-step workflows with more autonomy.