OpenAIGPT-6.1

GPT-6.1 Terra

GPT-6.1 Terra combines visual reasoning, screen interaction and writing. Unlike Astra and Sol, it offers a none reasoning setting, which makes the effort control relevant even for simple tasks.

Cache retention affects the write price

The published Standard cache-write rate is $1.25 per million tokens for 30 minutes of retention and $2.50 for 12 hours, before long-context adjustments. A cache read and a cache write are different billable operations.GPT-6.1 Terra model reference

Seeing an image does not mean generating one

Terra’s direct image support is input-only. It has no direct audio or video modality. A workflow that needs an image result must use a separate generation tool or model.GPT-6.1 Terra model reference

Ask for evidence in the image

Practical guidance. For a product-photo matching task, request the candidate item, visible distinguishing features and any ambiguity. Use a separate outcome for insufficient evidence. This makes an incorrect match easier to spot than a confident answer with no visual explanation.

Specifications & access

GPT-6.1 Terra specifications
DetailValue and conditions
AccessAPI
Model identifiersgpt-6.1-terra
Context window1,050,000 tokens

API context limit

GPT-6.1 Terra model reference
Maximum output128,000 tokens

Maximum output

GPT-6.1 Terra model reference
Input price1 USD per million tokens

Standard text input; up to 272,000 input tokens

GPT-6.1 Terra model reference
Output price5 USD per million tokens

Standard text output; up to 272,000 input tokens

GPT-6.1 Terra model reference
Input and outputText and images in; text outGPT-6.1 Terra model reference
Tool callsUse the Responses API.GPT-6.1 Terra model reference
Reasoning controlsnone, low, medium, high, xhigh and max; medium is the default.GPT-6.1 Terra model reference
Maximum input922,000 tokens within the total context window.GPT-6.1 Terra model reference

Related models

  • A lower-cost model for focused tasks at high volume.

    Launch date not confirmedAPI
  • A model for coding, tool use and complex professional tasks.

    Launch date not confirmedAPI