Gemini 3.1 Pro Preview is Google’s preview-tier Gemini reasoning model for demanding multimodal work, large-context analysis and sophisticated agent workflows. It is for teams prepared to use a preview API model in exchange for Google’s newest Pro-class capabilities.
At a glance
| Fact | Detail |
| --- | --- |
| Provider | Google |
| Type | Commercial preview |
| Best for | Coding, agents, research, multimodal analysis |
| Context window | 1,048,576 tokens |
| API available | Yes — Gemini API and Vertex AI |
| Local use | No |
| Last checked | 17 July 2026 |
Our verdict
Gemini 3.1 Pro Preview is the Google model to test when a million-token context window and broad multimodal capability matter more than release stability. It is a strong fit for research-heavy agents, codebases and document collections, but the preview label means production teams should keep a fallback model and watch Google’s release notes closely.
| Score | Rating |
| --- | --- |
| Overall | ★★★★☆ |
| Coding | ★★★★☆ |
| Agents | ★★★★☆ |
| Research | ★★★★★ |
| Budget | ★★★☆☆ |
Pricing
Standard Gemini API pricing is listed per million tokens.
| Meter | Price |
| --- | --- |
| Input | $2.00 / million tokens |
| Cached input | $0.20 / million tokens |
| Output | $12.00 / million tokens |
Google’s pricing can vary by product surface, region, context length and preview status. Check the live pricing table before committing a high-volume production workload.
Best for
- Long-document and codebase analysis
- Research agents that need large working memory
- Multimodal workflows spanning text, images and files
- Google Cloud and Vertex AI teams
Avoid when
- You need a stable, generally available model contract
- You need local inference or open weights
- You are running simple high-volume classification where a Flash-tier model is cheaper
What changed
- 17 July 2026: Profile created for Gemini 3.1 Pro Preview.
- 17 July 2026: Listed preview pricing and million-token context window checked against Google’s developer documentation.
- Future capability, availability and pricing changes will be added here.
Related news
- Google puts a personal face on workplace AI video
- Google faces fresh publisher challenge over Gemini training
- Alphabet jumps 4% on Dow debut as AI spending drains cash and talent
