What happened?
Google has announced Gemini 3.7 Flash, a new model version focused on coding and agentic workflows. The model arrived just three weeks after the release of Gemini 3.6 Flash; Google attributes this rapid update to developer feedback and algorithmic improvements.
Google is offering an introductory pricing plan for Gemini 3.7 Flash that will remain in effect through the end of 2026: $0.75 per million input tokens and $3.75 per million output tokens. Context caching is offered at $0.075 per million tokens. Starting January 1, 2027, standard prices will rise to $1.50 for input, $7.50 for output, and $0.15 for caching.
Why does it matter?
Google states that Gemini 3.7 Flash puts more effort into multi-step planning and tool calling, requiring fewer retries and less manual oversight. According to the company, the model recovers better from errors and executes multi-step plans more reliably.
This approach differs from Gemini 3.6 Flash's strategy, which focused on minimizing reasoning steps and tool calls; 3.7 Flash prioritizes improving execution quality instead. Google DeepMind said the new model shows improvements in debugging, generating functional web interfaces, and reasoning through real-world workflows.
What we know
- On the FrontierCode 1.1 Main benchmark, Gemini 3.7 Flash scored 43.6%, above Gemini 3.6 Flash's 34.4%.
- On DeepSWE v1.1, the model scored 65.3%, surpassing the previous version's 49.0%, though GPT-5.6 Terra remained ahead at 69.6%.
- On Code Arena, Gemini 3.7 Flash scored 1,588 Elo points; Claude Sonnet 5 scored 1,541, and GPT-5.6 Terra scored 1,523.
- On AutomationBench, the model scored 30.4%, outperforming Claude Sonnet 5's 10.7% and GPT-5.6 Terra's 23.6%.
- On Terminal-bench 2.1, Gemini 3.7 Flash scored 85.8% while GPT-5.6 Terra stayed ahead at 87.4%.
Comparison: Pricing
| Model | Input (per million tokens) | Output (per million tokens) |
|---|---|---|
| Gemini 3.7 Flash (introductory) | $0.75 | $3.75 |
| Gemini 3.7 Flash (post-2027) | $1.50 | $7.50 |
| Gemini 3.6 Flash | $1.50 | $7.50 |
| Claude Sonnet 5 | $2 | $10 |
| GPT-5.6 Terra | $2 | $12 |
What's next?
Gemini 3.7 Flash is being integrated into Gemini Spark, the personal AI agent for Google AI Pro and Ultra subscribers, and is also being made available for enterprise use through the Gemini Enterprise Agent Platform and the Gemini Enterprise app. Meanwhile, no release date has been announced for Gemini 3.5 Pro, Google's anticipated flagship model; Reuters and Axios reported that the model is undergoing joint testing, though the timeline remains uncertain.
For enterprise teams, the real deciding metric won't be price per token but cost per completed task. Whether Gemini 3.7 Flash's low introductory pricing and claimed accuracy gains hold up together in production environments will be tested in the coming months.