Gemini 3.7 Flash: Enhanced Model for Coding and Development
The Gemini team is excited to announce Gemini 3.7 Flash, our most advanced model yet for coding and agent tasks, building on the success of our Flash series. Released just three weeks after its predecessor, Gemini 3.6 Flash, this update is driven by valuable developer feedback and cutting-edge algorithmic enhancements that will influence future models. Notably, Gemini 3.7 Flash is available at a reduced introductory price, half that of the original 3.6 Flash, costing $0.75 per million input tokens and $3.75 per million output tokens.
Gemini 3.7 Flash offers significant improvements in several areas, including software engineering, knowledge work, and web development workflows. It excels in tasks such as debugging and issue resolution, with higher first-pass code accuracy and enhanced ability to generate production-ready code. This is demonstrated by its improved performance over 3.6 Flash in tests like FrontierCode 1.1 Main and DeepSWE v1.1.
In web development, 3.7 Flash creates more functional layouts and complete apps with fewer prompts. For UI generation, the model adheres closely to design references, whether they are screenshots, images, or full design systems. This version outperforms its predecessor in the WebDev Arena with a higher Elo score.
For fields rich in knowledge such as finance, law, and biosciences, Gemini 3.7 Flash enhances reasoning and accuracy. It achieves better results than 3.6 Flash on benchmarks like GDP.pdf, which evaluates how well a model can process complex documents, and AutomationBench, where it effectively completes real-world business tasks.
Using Gemini 3.7 Flash, developers can transform simple text prompts into fully playable 3D games and create interactive landing pages in a single attempt. The model facilitates faster learning for robotics and converts static PDFs into dynamic data stories with live charts and insights. Its improved developer experience means it navigates roadblocks more effectively, clarifies instructions, and follows directions with increased fidelity, reducing the need for manual interventions.
Gemini Spark, available to Google AI Pro and Ultra subscribers in over 160 countries, will now incorporate Gemini 3.7 Flash. This upgrade enhances Spark’s efficiency in handling knowledge work and using tools within Google Workspace, ensuring higher accuracy and quality in complex, multi-skill workflows.
Additionally, we are committed to strengthening Frontier Safety safeguards with Gemini 3.7 Flash, which includes updated protections against misuse in critical domains like Chemical, Biological, Radiological, and Nuclear (CBRN), as well as cyber threats, aligning with our bioresilience and cyber programs.
For more details, refer to the 3.7 Flash model card. To keep updated on our products, events, and offers, sign up for our newsletter. Rest assured, your information is handled in accordance with Google’s privacy policy, and you can opt out at any time.
