Today, Google is building on its popular Flash series with the launch of Gemini 3.7 Flash — its most intelligent workhorse model yet for coding and agents.
Takeaways 🚀
✅Gemini 3.7 Flash boosts coding
✅Stronger performance for AI agents
✅Faster workflows at lower cost
✅Better reasoning across complex tasks
✅Improved developer experience and accuracy
This release comes just three weeks after Gemini 3.6 Flash, and it’s a direct result of developer feedback and fresh algorithmic breakthroughs. Even better, Google plans to bring these advances to future models too. Overall, Gemini 3.7 Flash delivers big improvements across software engineering, knowledge work, and web development. And best of all, the introductory price is half of what the original 3.6 Flash cost per million tokens.
Better Intelligence for Complex Workflows
First and foremost, Gemini 3.7 Flash makes strong gains over 3.6 Flash on coding tasks like debugging and issue resolution. Not only does it get more code right on the first try, but it also produces more production-ready code. For example, on FrontierCode 1.1 Main, it scored 43.6% versus 34.4%. Similarly, on DeepSWE v1.1, it reached 65.3% versus 49.0%.
When it comes to web development, the model builds more functional layouts and feature-complete apps in fewer prompts. Moreover, when you generate a user interface, it closely matches your reference input — whether that’s a screenshot, an image, or a full design system. As a result, it beat 3.6 Flash on Arena.ai’s WebDev Arena with an Elo score of 1588 versus 1538.
Beyond coding, the model also shines in knowledge-heavy fields like finance, law, and biosciences. In those areas, its reasoning and accuracy improved noticeably. On the GDP.pdf benchmark, which tests how well a model handles complex documents, it scored 34.0% versus 22.0%. Likewise, on AutomationBench, it proved it can complete real-world business workflows more effectively, scoring 30.4% versus 17.0%.
To illustrate what’s possible, here’s what people are already building with it:
- A playable 3D game from a simple text prompt. By pairing Gemini 3.7 Flash with Nano Banana, developers generated characters, items, and textures in real time.
- Interactive landing pages in a single shot. In this case, Gemini 3.7 Flash orchestrated sub-agents and used Gemini Omni to create smooth parallax components.
- Faster robot training. Thanks to multimodal understanding in a three-agent graph loop, a robotics model learned significantly faster.
- Interactive data stories. As a result, complex annual reports transformed into engaging web experiences, complete with live charts and aggregated insights.
Better Developer Experience and Price
In terms of the developer experience, Gemini 3.7 Flash is noticeably smoother than 3.6 Flash. It adapts better to roadblocks, asks for clarity when needed, and follows instructions more faithfully.
On top of that, the model thinks more carefully. It puts more effort into multi-step planning and tool calls. Consequently, that disciplined approach means you spend less time on manual oversight and encounter fewer retries across your engineering workflows.
As for pricing, you can use Gemini 3.7 Flash through the end of the year at an introductory price of $0.75 per million input tokens and $3.75 per million output tokens. Together, this lower price and stronger performance let developers and customers scale production-ready agents without breaking the budget.
Early customer feedback, meanwhile, highlights the model’s performance and precision. Teams consistently report results far better than 3.6 Flash — and at a low cost.
Note: Introductory pricing ends on December 31, 2026. Starting January 1, 2027, prices will rise to $1.50 per million input tokens and $7.50 per million output tokens.
Improving Gemini Spark with 3.7 Flash
In addition to its API availability, Gemini Spark now runs on Gemini 3.7 Flash, starting today. Spark is available to Google AI Pro and Ultra subscribers in over 160 countries.
Originally launched at I/O as your personal AI agent, Spark runs 24/7 and takes action on your behalf — always under your direction. With this update, Spark becomes more efficient for knowledge work. It uses Google Workspace tools more effectively, so you get more accurate results and higher-quality output on complex, multi-skill workflows. In practice, Spark can now turn ideas into action more smoothly by consolidating files, drafting emails, and updating status documents.
Built with Safety in Mind
Throughout its development, safety has remained a top priority. Google continues to improve the coverage and strength of its Frontier Safety safeguards. As a result, Gemini 3.7 Flash ships with updated protections against misuse, focusing on Chemical, Biological, Radiological, and Nuclear (CBRN) risks, as well as cyber offense. At the same time, these safeguards still support beneficial use cases, in line with Google’s approach to bioresilience and its cyber program.
For more details, check the 3.7 Flash model card.
Try It Today
Whether you’re a developer, an enterprise, or an individual user, there’s a way to get started right now:
- Developers: Explore agent-first workflows in Google Antigravity, or start building in the Gemini API through Google AI Studio and Android Studio. From there, our developer guide will help you hit the ground running.
- Enterprises: Access 3.7 Flash directly in the Gemini Enterprise Agent Platform and the Gemini Enterprise app.
- Individuals: Use it through Spark, your 24/7 personal agent in the Gemini app, if you’re a Google AI Pro or Ultra subscriber in a supported country.
