Image: OpenAI

OpenAI has released GPT-6.1 Terra, a new model built to look at a screen and then do things on it, like clicking through apps, filling out forms and editing files. The company's pitch is simple. It wants Terra to handle a lot of everyday office work for far less money than its bigger models. OpenAI announced the model on X and in a blog post on October 26.

GPT-6.1 Terra: delivering fast and capable computer use at a lower price.

GPT-6.1 Terra understands what’s on screen, takes action across apps, writes naturally, and is a more capable model for your whole workday with the efficiency to take on more of it.

— OpenAI (@OpenAI) October 26, 2026

Terra is the cheaper, faster sibling in OpenAI's GPT-6 lineup, which also includes GPT-6.1 Sol and the top-end GPT-6 Astra. According to OpenAI, Terra's standard API prices are 50% lower than both GPT-5.6 Terra and GPT-6.1 Sol. The developer page lists $1 per million input tokens and $5 per million output tokens, with a context window of about 1.05 million tokens. OpenAI's own example of the kind of job it has in mind is reading incoming invoices and typing the details into accounting software.

The headline claims are about computer use. On Agents' Last Exam, a benchmark of professional tasks such as turning architectural drawings into a 3D model, OpenAI says Terra scores 15 percentage points higher than GPT-5.6 Terra at medium effort for about a third of the cost. The company also says it performs on par with GPT-6 Astra at one-tenth the price per token, and finishes tasks over three times faster than Astra and Sol.

OpenAI went after Anthropic directly too. At maximum effort, it says Terra is over three times as token efficient as Claude Opus 5.5 and costs a quarter as much per token, which works out to roughly nine times lower cost per task. These are OpenAI's own numbers and have not been independently checked at the time of this writing.

There are gains outside computer use as well. On an internal test built to trip models into factual mistakes, OpenAI says the error rate at low effort dropped from 53.6% with GPT-5.6 Terra to 16.7%. On Terminal-Bench 4.0, a coding test run from the command line, Terra solved 50.3% of tasks at max effort, up from 21.5% for its predecessor. OpenAI still says Sol and Astra are the better picks for the hardest coding work.

Safety is the part worth watching. In its system card addendum, OpenAI says it treats Terra as Critical in cybersecurity and High for biological and chemical capability under its Preparedness Framework. Because of that, Terra runs with the same safeguards used for GPT-6.1 Sol and GPT-6 Astra.

OpenAI said in a follow-up post that Terra is rolling out to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, and to developers as gpt-6.1-terra. OpenAI's ChatGPT release notes also list Business plans and say the model is not available in regular ChatGPT Chat. Users pick it from the model list under Advanced, and computer use has its own setup steps.

The Verge summed it up by noting that OpenAI says Terra is very good at computer use. How well that holds up on messy real-world screens is something early users will find out over the coming weeks.