OpenAI's GPT-6.1 Terra adds computer use at $1 per million input tokens

On Oct. 26, 2026, OpenAI released GPT-6.1 Terra. Its API model page lists $1 per million input tokens, $0.05 per million cached input tokens and $5 per million output tokens, with Batch and Flex 50% lower and Fast mode at 2x. The model supports a computer_use tool in the Responses API, so it can operate a browser or desktop screen by clicking and typing. OpenAI's release notes say it is offered to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, not in standard ChatGPT chat. We could read the API page but not the launch post, so OpenAI's performance claims are not checked here.

Computer use means an AI works a screen the way a person does. It matters where your tools have no integration. Take a 15-person plumbing shop where the office manager spends an assumed 6 hours a week copying job details from email into the field-service app and then into QuickBooks. At an assumed $25 an hour that is $150 a week, about $7,800 a year. If an agent reliably handled half, the shop would save about $3,900 a year. Token cost for that work is small: at the listed rates, a task using an assumed 200,000 input and 20,000 output tokens costs about $0.30. The real cost is checking its work and keeping it away from live financial data until it has proven itself. Care about this if your office re-keys data between systems. Ignore it if Zapier or a native integration already moves that data, or if nobody on your team pays for ChatGPT Plus or higher.