Gemini 3.7 Flash - Google DeepMind releases its latest flagship model.

What is Gemini 3.7 Flash?

Gemini 3.7 Flash is Google DeepMind's next-generation flagship AI model. Designed specifically for coding and agent workflows, it delivers significant leaps in benchmarks for software engineering, web development, and enterprise automation, surpassing Claude Sonnet 5 and GPT-5.6 Terra in some metrics. The initial price is only half that of the original 3.6 Flash, positioning it as a "cost-effective killer" for high-frequency agents and enterprise workflows.

Main functions of Gemini 3.7 Flash

• Software EngineeringIt supports the generation of high-quality production-grade code, with significant improvements over FrontierCode and DeepSWE benchmarks, and a marked enhancement in first-pass accuracy and debugging capabilities.
• Web DevelopmentSupports input from screenshots, images, or the complete Design System, generating high-fidelity interactive web pages with a single click; its Code Arena Elo rating is among the best in the market.
• Agent automationAutomationBench significantly improves performance when executing complex, multi-step enterprise workflows, automatically planning, calling tools, and adaptively adjusting when encountering obstacles.
• Knowledge-intensive jobsIt significantly enhances its ability to handle complex document reasoning in fields such as finance, law, and bioscience, and its comprehension of long documents such as GDP.pdf is also greatly improved.
• Multimodal creationWhen used with Nano Banana, it can transform text prompts into playable 3D games in real time, or convert static PDFs into dynamic web pages with interactive charts.
• Gemini Spark IntegrationAs a backend model for 24/7 personal agents, it drives automated operations such as document organization, email drafting, and status updates within Google Workspace.

The technical principles of Gemini 3.7 Flash

• Algorithm Iteration and Rapid Evolution MechanismThe model features post-training optimization and capability-oriented enhancement for Coding and Agent scenarios, rather than simply relying on pre-training scale expansion. This allows for a leapfrog improvement in the quality of software engineering and knowledge work while maintaining the speed advantage of the Flash series.
• Multi-step planning and rigorous reasoning executionThe model has a built-in more proactive inference resource allocation mechanism. When faced with complex tasks, it will invest more computing resources in multi-step planning and tool calls. When encountering obstacles during execution, it will adaptively adjust its strategy and clarify its intent. This "more rigorous execution" greatly reduces the need for manual supervision and the number of repeated retries, which is the core technical support for the leap in its agent and enterprise automation capabilities.
• Native Multimodal Unified Understanding3.7 Flash has a unified understanding capability across text, audio, images, code and video, enabling it to integrate visual input (such as screenshots and design drafts) with language instructions in scenarios such as 3D game generation, robot training loops and intelligent document conversion, achieving end-to-end generation from static PDFs to interactive web pages, and from text descriptions to runnable 3D assets.
• Agent Workflow and Sub-Agent Orchestration: It performs underlying optimizations for high-frequency agent scenarios, and supports the main model to schedule multiple sub-agents to complete tasks collaboratively through improved instruction following accuracy and roadblock adaptability; its orchestration capabilities enable it to process layout generation, interaction logic writing and visual component rendering in parallel in scenarios such as web development, making it an enterprise automation backend infrastructure that supports long-term online and large-scale calls.

How to use Gemini 3.7 Flash

• Developer accessDevelopers can directly access the model via the Gemini API or Google AI Studio.
• Quick ExperienceIn Google AI Studio, you can create projects at zero cost and choose the gemini-3.7-flash model to quickly test text, code, and multimodal capabilities.
• Mobile integrationAndroid Studio has integrated this model, allowing developers to access code assistance and generation features throughout the mobile application development process.
• Agent orchestrationGoogle Antigravity, an agent-first development environment, allows you to orchestrate sub-agents to build complex automated workflows.
• Enterprise deploymentEnterprise users can customize business processes and tool integration through the Gemini Enterprise Agent Platform to achieve large-scale, high-frequency calls.
• Enterprise useEnterprise employees can also directly access the Gemini Enterprise App through a unified portal to process documents and automate tasks using the built-in 3.7 Flash capabilities.

The core advantages of Gemini 3.7 Flash

• Extreme Iteration RhythmThe Flash series evolved from version 3.5 to 3.7 in just 3 months, and from 3.6 to 3.7 it was compressed to 3 weeks. Google's response speed to developer feedback and market competition is unparalleled among mainstream models.
• Coding ability leapFrontierCode 1.1 Main improved from 34.4% to 43.6%, and DeepSWE v1.1 jumped from 49% to 65.3%. First-pass code accuracy and long-cycle software engineering capabilities have been significantly enhanced, and the code output quality of the production-grade main model has been achieved.
• Agents and enterprise automation are absolutely leading.With an AutomationBench score of 30.4%, it not only nearly doubled its predecessor but also significantly outperformed Claude Sonnet 5 (10.7%) and GPT-5.6 Terra (23.6%), making it the strongest model for current enterprise workflow automation scenarios.
• Some benchmarks surpass flagship competitorsOn two key coding benchmarks, FrontierCode 1.1 (43.6%) and WebDev Code Arena (1588 Elo), Flash 3.7 has surpassed Claude Sonnet 5 and GPT-5.6 Terra, breaking the stereotype that "Flash is only used for lightweight auxiliary tasks".
• Ultimate cost-effectiveness barrierThe promotional price is $0.75 for input and $3.75 for output per million tokens, which is only half the original price of 3.6 Flash, and one-third to one-quarter of the price of Claude ($2/$10) and GPT ($2/$12). In high-frequency agents and enterprise-level workflows, this cost advantage is amplified exponentially with the volume of calls.
• Native multimodal end-to-end creationIt supports the direct generation of playable 3D games from text, the one-click conversion of screenshots/Design System into interactive web pages, and the automatic transformation of static PDFs into dynamic data stories with real-time charts, providing a complete creation process without the need for multi-model stitching.

Project address for Gemini 3.7 Flash

• Project official website：https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-3-7-flash/

Comparison of Gemini 3.7 Flash with similar competing products

Application scenarios of Gemini 3.7 Flash

• Intelligent Coding AssistantIt automatically generates production-grade code, debugs complex bugs, and completes long-cycle software engineering tasks, with a significant improvement in first-pass accuracy compared to the previous generation.
• Interactive Web DevelopmentGenerate high-fidelity interactive web pages with a single click from screenshots or Design System input, supporting multi-agent collaboration and parallax animation.
• Enterprise workflow automationIt drives RPA and business systems to automatically complete high-frequency repetitive tasks such as cross-platform data entry, approval workflow, and report generation.
• 3D Games and Content GenerationIt works with Nano Banana to transform text descriptions into playable 3D games in real time, dynamically generating characters, items, and textures.
• Intelligent document conversionAutomatically convert static PDF annual reports and research reports into dynamic web pages or data stories with real-time charts, data summaries, and interactive features.