OpenAI releases GPT-6.1 Terra for computer use and visual reasoning work

According to a BleepingComputer report by Mayank Parmar published on 26 October 2026, OpenAI has made GPT-6.1 Terra available. OpenAI pitches it as a quicker, lower-cost choice than its larger models for computer use, visual reasoning, coding and day-to-day work. The examples given range from form filling and video editing to reading technical drawings and automating routine desktop chores.

• The model is available to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, and in the API as gpt-6.1-terra.
• API pricing is 1 dollar per million input tokens and 5 dollars per million output tokens, which OpenAI says is 50 percent lower than GPT-5.6 Terra and GPT-6.1 Sol.
• OpenAI's internal benchmark puts the error rate at low reasoning effort at 16.7 percent, compared with 53.6 percent for GPT-5.6 Terra.
• It scored 50.3 percent on Terminal-Bench 4.0 at maximum reasoning effort; for the hardest coding work, OpenAI continues to point users to GPT-6.1 Sol or GPT-6 Astra.

Why it matters

Entering data on supplier portals, pulling measurements from drawings or repeating the same screen steps every day are examples of the routine work this model targets. The lower price and faster task completion OpenAI describes could lower the cost of trying AI agents on this kind of work. Keep in mind that the error rates come from OpenAI's own test sets and that agents working on sensitive systems should run with human approval; starting with a low-risk, highly repetitive process is a sensible first step.