# Gemini 3.7 Flash

Google's workhorse model for coding and agents. Gemini 3.7 Flash pairs a 1M tokens context window with text, image, audio, and video input, and lets you configure how much the model thinks per request.

- **Model ID:** `google/gemini-3.7-flash`
- **Type:** chat
- **Providers:** google, vertex
- **Context window:** 1,000,000
- **Maximum output tokens:** 65,536
- **Pricing:** $0.75/1M input tokens, $3.75/1M output tokens
- **Canonical page:** https://vercel.com/ai-gateway/models/gemini-3.7-flash

## Supported parameters

Detailed capability metadata has not been reported for this model.

## Example

```ts
import { streamText } from 'ai'

const result = streamText({
  model: 'google/gemini-3.7-flash',
  prompt: 'Why is the sky blue?'
})
```

## About

Gemini 3.7 Flash was released August 13, 2026 as Google's workhorse model for coding and agents, arriving three weeks after Gemini 3.6 Flash. It refines the reasoning foundation of 3.6 Flash rather than starting from a new pretraining run, and the improvements land mainly in coding and agentic execution.

Gemini 3.7 Flash accepts text, images, audio, and video, returns text, and works within a context window of 1M tokens with up to 65.5K tokens per response. Thinking is configurable per request, so you can spend more tokens on a hard problem and fewer on a routine one. The knowledge cutoff is March 2026.

Coding and agentic benchmarks moved the most between generations. Gemini 3.7 Flash scores 65.3% on DeepSWE v1.1, up from 49.0%, and 43.6% on FrontierCode 1.1 Main, up from 34.4%. Document comprehension on GDP.pdf reaches 34%, and long-context retrieval on GDM-MRCR v2 at 128k reaches 97.0%, which matters if you intend to actually fill the context window rather than just have it available.

Pricing is introductory and set to rise at the end of 2026. See the pricing panel on this page for current rates.

You can integrate Gemini 3.7 Flash through AI SDK, Chat Completions API, Responses API, Messages API, or other API formats, from TypeScript or Python. Routing rules move traffic from another Gemini model to Gemini 3.7 Flash without changing application code.

## What to consider

Current pricing is introductory and expires at the end of 2026, after which the rate roughly doubles. Anything you size on today's numbers should be re-checked against the pricing panel on this page before it becomes a long-term commitment.

Gemini 3.7 Flash is a refinement of Gemini 3.6 Flash rather than a new foundation, so the gains are concentrated in coding and agentic work. On workloads outside those areas the difference from 3.6 Flash may be small enough not to justify a migration.

The knowledge cutoff is March 2026, so pair Gemini 3.7 Flash with web search or retrieval for anything more recent. Thinking is configurable, which means an unconfigured request can spend more output tokens than you expect on a simple task.

## When to use

### Best For

- **High-volume coding work** where a workhorse rate matters more than frontier-tier reasoning
- **Long-context retrieval** that actually fills the 1M tokens window and needs accurate recall
- **Multimodal input** spanning text, images, audio, and video in a single request
- **Configurable thinking** that spends more tokens on hard prompts and fewer on routine ones
- **Agentic execution** on tasks that improved most between 3.6 Flash and this release

### Consider Alternatives When

- **Frontier reasoning** is the requirement, where a Pro-tier model handles the hardest problems
- **Long-term price stability** matters, since the introductory rate expires at the end of 2026
- **Non-coding workloads** dominate, where the gains over Gemini 3.6 Flash are smaller
- **Recent facts** are needed beyond the March 2026 cutoff without web search or retrieval

## Best for

- **High-Volume Coding:** Workhorse pricing on software work that runs constantly
- **Long-Context Retrieval:** Accurate recall across a filled 1M tokens window
- **Multimodal Input:** Text, images, audio, and video in one request
- **Configurable Thinking:** Token spend tuned per request instead of a fixed budget
- **Agentic Execution:** The area that improved most over Gemini 3.6 Flash

## Consider alternatives

- **Frontier Reasoning:** A Pro-tier model handles the hardest problems
- **Long-Term Price Stability:** The introductory rate expires at the end of 2026
- **Non-Coding Workloads:** Gains over Gemini 3.6 Flash are smaller outside coding
- **Post-Cutoff Facts:** March 2026 knowledge needs web search or retrieval

## Frequently asked questions

### What is Gemini 3.7 Flash built for?

Coding and agentic work at workhorse pricing. Google positions it as its most intelligent model in that tier rather than a frontier reasoning model.

### How does Gemini 3.7 Flash differ from Gemini 3.6 Flash?

It refines the same reasoning foundation rather than starting from a new pretraining run. Coding and agentic benchmarks moved most: DeepSWE v1.1 from 49.0% to 65.3% and FrontierCode 1.1 Main from 34.4% to 43.6%.

### What input types does Gemini 3.7 Flash accept?

Text, images, audio, and video. It returns text.

### What is the context window for Gemini 3.7 Flash?

The context window is 1M tokens, with up to 65.5K tokens per response. Long-context retrieval on GDM-MRCR v2 at 128k reaches 97.0%.

### Will the price of Gemini 3.7 Flash change?

Yes. Launch pricing is introductory and expires at the end of 2026, after which the rate roughly doubles. Check the pricing panel on this page for current rates.

### Can I control how much Gemini 3.7 Flash thinks?

Yes. Thinking is configurable per request, so you can raise the budget for hard problems and lower it for routine ones. An unconfigured request may spend more output tokens than a simple task warrants.

### What is the knowledge cutoff for Gemini 3.7 Flash?

March 2026. Pair Gemini 3.7 Flash with web search or retrieval for facts that changed after that date.

### Can I move traffic to Gemini 3.7 Flash without changing my code?

Yes. Add an AI Gateway routing rule that rewrites requests from another Gemini model to `google/gemini-3.7-flash`. Your application keeps sending the old model identifier.

### Does Gemini 3.7 Flash support Zero Data Retention?

Yes, Zero Data Retention is available for this model. Zero Data Retention is offered on a per-provider basis. See https://vercel.com/docs/ai-gateway/capabilities/zdr for details.

## Links

- [Model page](https://vercel.com/ai-gateway/models/gemini-3.7-flash)
- [AI Gateway documentation](https://vercel.com/docs/ai-gateway)
- [Provider model documentation](https://deepmind.google/models/model-cards/gemini-3-7-flash/)