OpenAI on Monday released GPT-6.1 Terra, a faster and cheaper AI model built for routine computer work. It is available from today to ChatGPT Plus, Pro, Enterprise and Edu users, and in Codex.
The launch gives OpenAI a new middle-tier model that can look at a screen, understand what it shows and act on it by clicking, typing and moving between apps. The company said Terra can read incoming invoices and enter the details into accounting software, fill in forms, draft proposals, read technical drawings and edit videos.
Pricing is the part that will matter to most developers. OpenAI said Terra costs $1 per million input tokens and $5 per million output tokens, half the price of GPT-5.6 Terra. In the company’s own testing, Terra scored 15 percentage points higher than the model it replaces while costing about a third as much, and it matched GPT-6 Astra at a tenth of the price.
OpenAI also said Terra is better at understanding images. In one internal test, it turned architectural drawings into a 3D model. In another, it watched an animated film shot by shot and produced a time-coded log for a restoration team. The company said Terra’s error rate at the lowest setting fell from 53.6% to 16.7% on questions designed to trip models up, and that it solved about half the tasks in a command-line coding test, more than double its predecessor’s score.
That still leaves a familiar problem: the most eye-catching comparisons come from OpenAI’s own testing, not independent checks. The company says Terra reaches the Critical level for cybersecurity skills under its Preparedness Framework and carries the same protections as GPT-6 Astra, but it also canceled GPT-6.1 Astra last month after it failed internal safety tests. OpenAI says Terra is more likely than its predecessor to stay within the task it is given and to flag its limits, yet it still recommends Sol or Astra for the hardest coding work.
For now, Terra looks designed to make higher-volume office automation cheaper, not to replace OpenAI’s most capable systems. The question is how far those savings travel once outside OpenAI’s own tests, where the model’s speed, safety claims and cost math will need to stand on their own.
