Google released Gemini 3.7 Flash on August 13, 2026, described by the company as its most capable Flash model, built for complex coding and agentic work. The model arrived directly as stable in the Gemini API documentation, and its pricing is identical to its predecessor.
Key takeaways
- Released August 13, 2026, marked as stable (New Stable) in the Gemini API docs
- Positioned for coding, agentic tasks and reliable multi-step execution
- Introductory price through December 31, 2026: 0.75 USD per million input tokens and 3.75 USD per million output
- From January 1, 2027 both rates double, to 1.50 USD and 7.50 USD per million tokens
- Three reasoning-effort levels: high, medium and low (the minimal option from 3.6 Flash was removed)
A model for code and agents
Google describes Gemini 3.7 Flash as its best and most capable Flash model, built for complex coding, agentic workflows?agentic workflows: Tasks where the model autonomously runs a longer sequence of steps and tool calls rather than giving a single answer. and reliable multi-step execution. It is a clear signal of where the Flash line is heading, toward models meant to run longer sequences of actions and tool calls on their own.
The model keeps three reasoning-effort levels, high, medium and low. Compared with Gemini 3.6 Flash, the minimal option is gone, suggesting Google is aiming at tasks that need real reasoning rather than only fast answers. The model remains multimodal and supports server-side tools, including code execution.
Pricing unchanged from the predecessor
Against expectations of a cut, Gemini 3.7 Flash pricing matches Gemini 3.6 Flash. During the introductory period, through December 31, 2026, the model costs 0.75 USD per million input tokens?tokens: The basic unit of text a model processes, a word fragment or characters. Model pricing is billed per million tokens. and 3.75 USD per million output. From January 1, 2027 both rates double, to 1.50 USD and 7.50 USD.
| Period | Input | Output |
|---|---|---|
| Through Dec 31, 2026 | 0.75 USD | 3.75 USD |
| From Jan 1, 2027 | 1.50 USD | 7.50 USD |
This is an important nuance. Google is raising the Flash model's capability without changing the entry price, which means a better capability-to-cost ratio for the same token budget.
Where it runs
The model is available in the Gemini API and in Google AI Studio as a stable version. Support also arrived quickly outside Google's ecosystem, the llm-gemini plugin version 0.33 added the gemini-3.7-flash identifier.
Why it matters
Holding the price while raising capability is a quiet but meaningful move. Competition in the cheap, fast model segment is no longer only about the lowest rate per token, but about how much real agentic work a model does for the same cost. By steering the Flash line toward coding and multi-step tasks, Google is targeting the fastest-growing use, autonomous agents running sequences of tool calls. That shifts the axis of competition from price to the quality of execution.
What's next
- The introductory price runs through December 31, 2026, from January 1, 2027 token rates double to 1.50 USD and 7.50 USD per million, per the Gemini API pricing
- The model is already stable in the Gemini API and Google AI Studio, so the next test is real adoption in agentic tools and comparisons with rival coding models
Sources
- Google AI for Developers, Gemini API pricing
- Google AI for Developers, Gemini API models
- Simon Willison's Weblog, llm-gemini 0.33
