GPT-6.1 Terra lets ChatGPT use your apps for you. Here is what to know first
OpenAI released GPT-6.1 Terra on Monday, October 26. This AI model can use a computer to do tasks for you. It looks at the screen, clicks buttons, types words and switches between apps to get things done. It is now becoming available to people who pay for ChatGPT. In OpenAI's own test, which used prompts designed to make the model fail, Terra still made mistakes about one time in six when using its fastest setting.
What is “computer use”?
Most chatbots answer questions. A model that can use a computer can also do things. You tell it what you want, such as copying details from invoices into accounting software or filling out a long form. The AI then uses apps much like a person does. It reads what is on the screen, clicks buttons and types into boxes.
Who can use it?
OpenAI says GPT-6.1 Terra is becoming available to ChatGPT Plus, Pro, Enterprise and Edu users through ChatGPT Work and Codex, OpenAI's coding tool. People with free accounts cannot use it. ChatGPT Plus is the cheapest plan that includes Terra and costs $20 per month. Developers can also use Terra through OpenAI's paid API, a service that lets them connect the model to their own software. OpenAI says Terra costs half as much as the model it replaces.
What can it do?
OpenAI says Terra can read invoices and enter their details into accounting software, fill out forms, write proposals, read technical drawings and edit videos. OpenAI says it completes these tasks more than three times faster than its larger models.
How often does it get things wrong?
In OpenAI's own test, which used prompts designed to cause mistakes, Terra gave a wrong answer 16.7% of the time at its lowest reasoning setting, which is also its fastest setting. The model it replaces had a mistake rate of 53.6%. When Terra used its most careful setting, its mistake rate dropped to 4.3%. This is a big improvement. But at the fastest setting, about one answer in six is still wrong. When an AI clicks a submit button for you, a wrong answer can lead to the wrong action. These numbers come from OpenAI's own tests and have not been checked by independent testers.
What should I think about before letting AI use my accounts?
- It can see what is on your screen. While the model is working, it can see information on your screen, including bank balances, messages and personal details.
- Mistakes can lead to real actions. You can ignore a wrong answer in a chat. But a wrong click could send a message, delete a file or make a payment.
- Look for a stop button. Check whether the settings make the AI ask you before sending, buying or deleting anything. Keep these safety checks turned on.
- Start with small tasks. Sorting files is a safer way to test the AI than asking it to pay bills.
Is it worth paying for?
If you run a small business and spend a lot of time entering the same kinds of information, Terra could save you many hours, as long as you check its work. For most people, the free version of ChatGPT is still enough for everyday questions. Being able to use a computer is not, by itself, a good reason to pay for an upgrade.
What to do next
- If you already pay for ChatGPT, try Terra on one simple task with little risk, and watch everything it does.
- Before giving it real work, turn on any settings that make it ask you before taking action.
- Never let it handle passwords, payments or anything you cannot undo.
Main source: OpenAI's announcement
Sources: OpenAI, “Introducing GPT-6.1 Terra,” October 26, 2026; OpenAI on X, October 26, 2026; BleepingComputer, October 26, 2026; The Next Web, October 26, 2026.
