Overview
Gemini 3.7 Flash is a high-speed flagship model launched by Google on August 13, 2026, continuing the Flash series' advantages in speed and cost, designed for high-throughput agent and programming scenarios. The model achieved a score of 65.3% on the DeepSWE v1.1 benchmark, a significant improvement over the previous generation 3.6 Flash, while maintaining low-latency response characteristics. As one of the core models in Google AI Studio and the Gemini API, Gemini 3.7 Flash offers highly competitive input and output unit prices at an introductory price, aiming to lower the barrier for developers to scale AI capabilities and promote the deployment of agent applications in real business scenarios.
Key Features
- High-Speed Inference Capability: Continuing the Flash series positioning, it optimizes inference latency, suitable for agent workflows requiring real-time responses and high-concurrency call scenarios.
- High Score on DeepSWE v1.1: Achieved a score of 65.3% on the software engineering benchmark DeepSWE v1.1, surpassing the previous generation 3.6 Flash, demonstrating stronger code understanding and generation capabilities.
- Low-Cost Pricing: During the introductory period, the price is $0.75 per million input tokens and $3.75 per million output tokens, providing a cost-effective solution for high-throughput tasks.
- Dual Platform Access: Available through Google AI Studio and the Gemini API, facilitating quick integration into existing applications or prototype validation for developers.
- Optimized for Agent Scenarios: Tuned for multi-step tasks, tool calling, and context management, enhancing the stability and execution efficiency of autonomous agents.
Use Cases
- Large-scale code generation and review to boost development team productivity
- High-concurrency intelligent customer service bots to handle massive user queries
- Automated test case generation and defect analysis
- Real-time text processing and classification in data pipelines
- Task scheduling and decision support in multi-agent collaboration systems
Pros
- High-speed response, suitable for real-time interaction scenarios
- Leading score on DeepSWE v1.1, outstanding programming capabilities
- Introductory price is highly attractive, reducing trial costs
- Official dual-platform support, flexible deployment
- Maintains the lightweight characteristics of the Flash series, resource-friendly
Pricing
The introductory pricing is $0.75 per million input tokens and $3.75 per million output tokens. This price is effective from August 13, 2026, until January 2027 when it reverts to the original price (doubled). Specific restoration dates and original price details are subject to official announcements from Google.
Summary
Gemini 3.7 Flash is Google's high-speed flagship model for high-throughput and programming scenarios, surpassing its predecessor with a DeepSWE v1.1 score of 65.3% and attracting developers through an introductory low-price strategy. Its core advantage lies in the balance between speed and cost, making it suitable for agent-type applications. Pricing will adjust in January 2027, and long-term costs should be monitored through official information.