GPT-6.1 Terra is OpenAI's computer-using model, at half the API price of its predecessor
The strangest part of OpenAI's newest model is not what it knows. It is what it does with your screen.
Quick take: GPT-6.1 Terra is a model built to operate computer apps directly. It looks at the screen, clicks buttons, types text, and moves between apps to finish multi-step tasks. OpenAI cut its standard API price in half compared with GPT-5.6 Terra, and it is available in ChatGPT Work and Codex for Plus, Pro, Enterprise, and Edu users, plus developers through the API.
What it actually does
Most chatbots answer you in text and stop there. Terra is meant to keep going: draft a proposal, read a technical drawing, fill in a form, edit a video. OpenAI chief Sam Altman announced it on X with a short 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."
The model does more than answer questions. It takes over the clicking and typing for boring computer work.
Why the price cut matters
Terra costs 1 dollar per million input tokens, 5 cents per million cached input tokens, and 5 dollars per million output tokens. OpenAI also says it finishes tasks about three times faster on average than its more advanced siblings, and scored 15 percentage points higher than its predecessor on an Agents' Last Exam test of computer use, at one-third of the cost per million tokens. Those are company numbers, so treat them as claims, not independent results.
Still, cheaper and faster changes who can afford to automate office work. A tool that was a luxury experiment becomes something a small team can run all day.
The catch worth knowing
A model that clicks buttons needs permission to reach your apps and files, and it can still get things wrong. OpenAI's own framing says to review important actions rather than assume every task will be completed correctly. So the skill that matters now is checking the work before anything gets sent, filed, or paid.
Bottom line
Terra turns the chatbot into a coworker that can use a mouse. The price cut means more people will try it, and the risk moves from bad answers to wrong clicks. Would you let an AI run through your apps to finish a boring task, or would you watch every single click?
Cover image: AI-generated