Gemini 3.1 Pro Preview
Google - gemini-3.1-pro-preview
8/10
Provisional 0 practitioner reviews Source verifiedContext window
1,048,576 tokens
Input price
$2/1M
Output price
$12/1M
Checked
May 16, 2026
Pricing caveat: Standard paid-tier pricing for prompts up to 200K tokens; prompts above 200K have higher input and output prices.
Decision Notes
Best for
- - Long-context multimodal RAG over documents, images, video, audio, or PDFs
- - Agentic coding tasks that benefit from Gemini tool support
- - Teams that can tolerate preview-model lifecycle risk for higher capability
Avoid when
- - Production systems that require stable model IDs only
- - Prompts above 200K tokens where Gemini Pro long-context pricing increases materially
Dimension Breakdown
Reliability of function/tool calling — correct schema adherence and parameter extraction
Price per token relative to output quality for agent tasks
Response time — time to first token and total generation time under load
Uptime, rate limit headroom, and error rates in production
Long-context coherence and instruction following over turns
Share Your Experience
Have production experience with Gemini 3.1 Pro Preview? Add the first real review to improve confidence in this score.
Submit a ReviewTop Use Cases
Summary
Google preview model with a 1,048,576-token input limit and strong multimodal agent features. Its preview status and tiered pricing deserve explicit production testing.
Sources
Practitioner Reviews
No reviews yet
Be the first to share your experience with Gemini 3.1 Pro Preview.
Related Models
- Gemini 3.1 Flash-Lite (Google, 8.6/10)
- GPT-5.4 mini (OpenAI, 8.6/10)
- Claude Opus 4.7 (Anthropic, 8.4/10)
Last updated: May 16, 2026