GPT-6.1 Terra: OpenAI's new model at half the API price, out in ChatGPT Work, Copilot and Bedrock

OpenAI released GPT-6.1 Terra on 26 October 2026. The company calls it a fast, efficient model with detailed visual reasoning, precise computer use and clearer writing. Its examples include drafting proposals, reading technical drawings, filling in forms and editing videos.

Standard API prices are $1 per million input tokens, $0.05 per million cached input tokens and $5 per million output tokens. OpenAI says that is 50% lower than GPT-5.6 Terra and GPT-6.1 Sol. Its example of the high-volume work this suits is reading incoming invoices and entering them into accounting software.

OpenAI's pricing page lists GPT-6.1 Sol at $2 input and $10 output per million tokens, and GPT-5.6 Terra at $2 and $12. GPT-6 Astra is $10 and $50. Batch and Flex processing halve Terra's standard price, to $0.50 and $2.50.

On OpenAI's own tests at medium effort, Terra beat GPT-5.6 Terra by 15 percentage points on a benchmark of professional tasks. It did so at about a third of the cost. OpenAI says it matches GPT-6 Astra on that test at a tenth of the price per token. On OpenAI's internal test of prompts built to cause factual errors, its error rate at low effort was 16.7%, against 53.6% for GPT-5.6 Terra.

OpenAI also says that for the most complex coding tasks, GPT-6.1 Sol and GPT-6 Astra remain the better choice.

The saving is largest on high-volume agent and document work, such as OpenAI's invoice example. Output tokens fall from $12 to $5 per million against GPT-5.6 Terra, which is 58% lower by our arithmetic.

Our own illustration: a team processes a million invoices a month, at an assumed 2,000 input tokens and 500 output tokens each. At standard prices, before caching or any regional uplift, that costs about $4,500 a month on Terra. The same work costs $9,000 on GPT-6.1 Sol and $10,000 on GPT-5.6 Terra.

The saving holds only if Terra does the job as well on your own documents. Every performance figure so far comes from OpenAI, or from GitHub's early testing in Copilot.

On OpenAI's advice, a team can likely move routine agent and document steps to Terra and leave the most complex coding on Sol or Astra.

Terra is available to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, and to developers as gpt-6.1-terra in the OpenAI API. OpenAI says a usage allowance covers more tasks with Terra than with GPT-5.6 Terra, and substantially more than with Sol or Astra.

OpenAI's Enterprise and Edu release notes say Terra is rolling out in Work and Codex, and is missing from Chat. Members can use it when they have permission for both the model and Work or Codex. Terra doesn't support UAE inference residency, and OpenAI says the limit depends on where processing runs, whatever the user's location.

GitHub made Terra generally available in Copilot on 26 October 2026, on the Pro, Pro+, Max, Business and Enterprise plans. GitHub bills it at the provider's list price under usage-based billing. On Business and Enterprise, new models switch on automatically unless an administrator has turned off the global default or disabled this model.

AWS made Terra generally available on Amazon Bedrock the same day, with up to 1 million input tokens of context and text and image inputs. Teams can call it through Bedrock's APIs, or set ChatGPT Work and Codex to use the model on Bedrock. AWS says its prices are at least 50% lower than GPT-6.1 Sol and GPT-5.6 Terra, without giving figures or regions.

In Copilot, Terra can appear without anyone choosing it. An organisation that kept GitHub's default on will find Terra in its developers' model list as the rollout reaches it, billed at list price.

For a ChatGPT Enterprise admin, access to Terra follows the model-access settings and assigned roles. Staff who work in Chat won't see Terra, so a rollout plan should point them to Work or Codex.

For Gulf buyers, a ChatGPT workspace that relies on UAE inference residency can't use Terra, and AWS's announcement names no Bedrock regions. A Gulf team should check both before planning on Terra.

Bedrock gives AWS customers another route to the same model under their existing AWS access and audit controls. The price on Bedrock still needs checking, because AWS gives only a percentage.