In This Article

OpenAI just filled the most important gap in its model lineup. On October 26, 2026, the company released GPT-6.1 Terra, a fast mid-tier model built for one job: doing work on a computer — reading screens, clicking buttons, filling in forms, and moving between apps, the way a human operator would.

The headline is price. Terra costs $1 per million input tokens and $5 per million output tokens, which OpenAI says is 50% lower than both the GPT-5.6 Terra it replaces and the pricier GPT-6.1 Sol. For companies automating back-office work at scale, that is the difference between a pilot project and everyday economics.

What is GPT-6.1 Terra?

Terra is the workhorse tier of OpenAI's four-model family: Luna at the bottom, Terra in the middle, then Sol and the flagship Astra at the top. The Terra line was the last one still running on the GPT-5.6 generation, so this release finally brings the whole lineup into the GPT-6.1 era.

OpenAI describes Terra as "a fast, efficient model that combines detailed visual reasoning and precise computer use alongside clearer, more natural writing." Its targets are the tasks that fill a workday: drafting proposals, interpreting technical drawings, filling out forms, editing videos — and, the example OpenAI keeps returning to, reading incoming invoices and typing the details into accounting software.

The model reads text and images as input and writes text back, with a context window of just over a million tokens. In the API it is called as gpt-6.1-terra.

What "computer use" actually means

"Computer use" is OpenAI's term for an AI that interprets a screen and takes actions inside it: clicks, typing, navigation between applications. It is a step beyond answering questions — the model does the task inside your existing software instead of describing how to do it.

OpenAI CEO Sam Altman announced the release on X with the line: "6.1 Terra is here! It is good at a lot of things, but we are especially excited for this new era of computer use."

This is also the release that signals a real industry shift: the fight is moving from chatbots to agents that act inside existing enterprise software. Terra is OpenAI's bet that the bulk of that work doesn't need a flagship-class model — it needs a fast, cheap one that reliably clicks the right button.

Benchmarks: what OpenAI claims

An honest framing first: every benchmark number below comes from OpenAI's own testing. Independent results are not out yet, so treat these as claims until third parties verify them.

  • Agents' Last Exam (professional computer tasks like turning architectural drawings into a 3D model): Terra at medium reasoning effort scores 15 percentage points higher than GPT-5.6 Terra at about one-third the cost, and matches GPT-6 Astra at a tenth of the price per token. OpenAI also says it finishes tasks more than 3x faster than Astra and Sol.
  • Terminal-Bench 4.0 (command-line coding agents): Terra solves 50.3% of tasks at maximum effort, up from 21.5% for GPT-5.6 Terra — more than double.
  • Factual reliability: on OpenAI's internal benchmark of error-inducing prompts, the error rate at low effort fell from 53.6% to 16.7% (and 4.3% at maximum effort).
  • Against Anthropic: OpenAI claims Terra at max effort is over 3x more token-efficient than Claude Opus 5.5 at a quarter of the price per token, working out to roughly 9x lower cost per task.

Also in the release: the same day, GitHub made GPT-6.1 Terra generally available in Copilot for everyday agentic coding and terminal work, cheaper than running GPT-6.1 Sol — and early testing showed it solved noticeably more terminal and C# tasks than GPT-5.6 Terra.

Pricing and availability

Input tokens$1 per million
Cached input$0.05 per million
Output tokens$5 per million

That is 50% below GPT-5.6 Terra (which cost $2 input / $12 output) and exactly half of GPT-6.1 Sol's rates — roughly a tenth of flagship Astra per token. OpenAI also says that in ChatGPT Work and Codex, the same plan allowance stretches further with Terra than with its predecessor.

Who gets it: ChatGPT Plus, Pro, Business, Enterprise, and Edu users in ChatGPT Work and Codex, plus API access for developers. It is also available through Amazon Bedrock. Free and Go plans still get the older GPT-5.6 Terra for now — OpenAI has not announced a free-tier rollout. Note that Terra is not available in Chat itself (it lives in Work and Codex), and organizations requiring inference inside the UAE are excluded.

Where Terra fits in the OpenAI lineup

  • GPT-6 Luna — the lightweight, cheapest option for simple everyday tasks.
  • GPT-6.1 Terra — the mid-range workhorse: computer use, visual reasoning, and everyday coding at half the old price.
  • GPT-6.1 Sol — the stronger default for complex reasoning and serious coding.
  • GPT-6 Astra — the flagship, for the hardest work.

OpenAI's own guidance: for the most complex coding tasks, Sol and Astra remain the better choice. Terra's pitch is volume — routine agent work that never penciled out at Sol prices. See our GPT-6.1 Sol vs GPT-6 Luna comparison for how the tiers differ, and our tested AI coding assistants roundup for tools you can pair it with.

Pros and cons

Pros

  • 50% cheaper API pricing than its predecessor — the most affordable serious agent model OpenAI has shipped.
  • Strong computer-use numbers on OpenAI's own tests, including a 15-point jump over GPT-5.6 Terra.
  • Error rate on tricky prompts dropped from 53.6% to 16.7% at low effort.
  • Already available in GitHub Copilot, ChatGPT Work, and Codex on day one.
  • Cached input at $0.05/M keeps repeat workflows cheap.

Cons

  • All benchmark figures are OpenAI's own — independent verification is still pending.
  • Not available in Chat or on the free tier; organizations needing UAE data residency are excluded.
  • Still not the pick for hard coding — OpenAI itself points complex work to Sol or Astra.
  • Regional data processing adds 10% to the price, and prompts longer than ~272K tokens cost double on input.

Verdict

GPT-6.1 Terra is not the flashy release of the month — but it may be the most useful one. Most businesses don't need a flagship model to read invoices into accounting software; they need one that does it reliably and cheaply enough to run all day. At $1/$5 per million tokens with a big jump in computer-use ability, Terra makes the economics of AI agents work for back-office automation.

If you automate routine screen-and-form workflows, Terra deserves the first slot in your evaluation — and if you want a second opinion on the numbers, wait for independent benchmarks before committing at scale.

FAQ

Is GPT-6.1 Terra available on the free ChatGPT plan?
No. Terra is in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise, and Edu users. Free and Go plans still use GPT-5.6 Terra for now.

Is GPT-6.1 Terra better than GPT-6.1 Sol?
Not for hard tasks. OpenAI says Sol (and Astra) remain better for complex reasoning and the most difficult coding. Terra wins on price, speed, and everyday computer-use work.

How much does GPT-6.1 Terra cost?
$1 per million input tokens, $0.05 per million cached input, and $5 per million output tokens — 50% lower than GPT-5.6 Terra and half of GPT-6.1 Sol's rates.

Can I use GPT-6.1 Terra in GitHub Copilot?
Yes. GitHub made GPT-6.1 Terra generally available in Copilot on launch day for Pro, Pro+, Max, Business, and Enterprise plans, aimed at everyday agentic coding and terminal work.

Further reading: OpenAI's broader model strategy is covered in our OpenAI Decisions API explainer and the ChatGPT Chrome extension guide — another way ChatGPT works inside your browser. Details on Terra's launch were cross-checked against DataCamp's breakdown and gHacks' report; all benchmark figures are OpenAI's own until independently verified.