GPT-6.1 Terra

What GPT-6.1 Terra is

GPT-6.1 Terra is a cloud API model positioned for visual understanding, computer use and writing. It was released on 26 October 2026. Understanding a screen and using a computer are related but separate: an application must supply the tools, environment and permissions that let a model act.

For a beginner, the important distinction is between generating an explanation and taking an action in software. A model can describe what it sees, while a computer-use system can make clicks or other changes. That second workflow needs explicit boundaries and a way to inspect the result.

How it fits into a workflow

An illustrative evaluation uses a harmless screenshot and asks the model to identify the next step in a clearly described task. Check that interpretation before allowing actions in a controlled environment. For writing work, provide the audience, source material and requested structure, then assess the resulting document rather than only the conversational answer.

Examples described as illustrative or proposed trials are editorial suggestions. Documented real builds, when available, are linked separately below.

Requirements, cost and limitations

Check API access, current prices and the tool environment's permission model. Computer-use support does not establish compatibility with every website or unrestricted control. OpenAI still positions Sol and Astra for its hardest coding workloads, so Terra's visual emphasis should not be turned into an overall ranking.

When to choose it and what to compare

Consider Terra for visual and writing-oriented tasks where those capabilities are central. Use a task-specific trial to decide whether computer use is necessary at all; a suitable API or ordinary document-processing route may be more direct. The directory has not benchmarked these alternatives.

Access, running cost and practical requirements

Frontier hosted candidate — editorial selection. How we use this label.

OpenAI API and eligible ChatGPT Work/Codex plans. Launch specifically lists Plus, Pro, Enterprise and Edu. A computer-use application still supplies the computer, tools and permissions.

Developer API prices

Billing route

Standard API

Units

USD per one million tokens

Input

$1

Cached input

$0.05

Output

$5

Above 272K input tokens, full-request input/cache rates double and output rate rises 50%. Batch/Flex half Standard; Fast 2x; regional processing can add 10%.

API prices are not a chat subscription price. A token is a unit of processed text; images and other inputs have their own token accounting. Repeated history, reasoning output and tools can change the final bill. A subscription may offer separately claimed credits, subject to its terms.

What the configuration changes

Official examples include matching product photographs, interpreting drawings, acting on application screens and revising text.

Supports none/low/medium/high/xhigh/max effort.

No native audio/video input documented. Tool calling requires Responses API.

Provider still recommends Sol/Astra for its hardest coding tasks; no universal winner follows.

Benchmark evidence and what it means

These are selected source-reported observations, not a league table. Task, version, tool access, reasoning budget and test harness all affect a result. A historical score does not establish current leadership or accuracy on your own work.

Sources and evidence

Product facts are based on the following published sources, checked 2026-10-27. Workflow suggestions and comparisons are editorial interpretation; PrisAIm has not independently benchmarked this product.