Google · Gemini
Gemini 3.1 Pro Preview
A preview model for precise multi-step reasoning, software engineering and tool use across text, images, video, audio and PDFs.
Use for evaluation, not quiet production dependency. Record the exact model ID and rerun tests before any upgrade.
Use it for
- Complex multimodal analysis
- Precise tool-using workflows
- Software engineering evaluation
Preview status makes it a poor default for workflows that require stable behaviour and long-term operational certainty.
Effort without the jargon
The task needs deliberate multi-step reasoning and preview volatility is acceptable.
Advanced recordCost, limits, privacy and operations
The specifications are substantial, but preview status is the bigger operational fact: pin the ID, expect change and do not build a critical workflow without rerunnable tests.
- API cost
- Depends on surfaceThe official Gemini API price table is tiered by modality and consumption mode; preview status and service route can affect limits and availability.
- Context window
- Officially disclosed1,048,576 input tokens
- Maximum output
- Officially disclosed65,536 output tokens
- Knowledge cutoff
- Not disclosedNot stated on the model card
- Recorded inputs
- Text · Images · Video · Audio · PDF
- Recorded output
- Text
Data handling is a product decision
Depends on surfaceFree and paid developer tiers and Vertex AI can have different data-use and governance conditions. Confirm the service and account, not only the model ID.
Operational constraints
- Preview models can change before stable release and may have more restrictive rate limits.
- The custom-tools endpoint can fluctuate on work that does not benefit from bash or custom tools.
- Audio generation, image generation, the Live API and computer use are not supported on this model record.
Unknown means the cited official record does not disclose a safe value. It is not an estimate. Prices are provider list prices in USD where stated and can change before this record's review date.
Keep the evidence separate
Google describes 3.1 Pro Preview as a more grounded, token-efficient Pro model for software engineering and reliable agentic execution.
No sufficiently useful independent note has been added yet. That absence is not filled with our guess.
No Human Bit result is claimed; use this as evaluation guidance only.
Not scheduled
No Human Bit result is claimed for this model yet.