OpenAI releases GPT-6.1 Terra at half the price

OpenAI has announced GPT-6.1 Terra, the first Terra-class model since GPT-5.6. OpenAI describes it as a fast, efficient model for computer use, visual reasoning and writing. Plus, Pro, Enterprise and Edu users can use it now in ChatGPT Work and Codex, and developers can call it in the OpenAI API as gpt-6.1-terra.

Standard API pricing is $1 per million input tokens and $5 per million output tokens. That is half or less of GPT-5.6 Terra's $2 and $12, and OpenAI says it is also 50% below GPT-6.1 Sol's price. Cached input costs $0.05 per million, down from $0.20. According to Artificial Analysis, cache writes cost half as much as before, at $1.25 per million.

OpenAI's own benchmarks focus on computer use and factuality. On Agents' Last Exam, Terra at medium effort scores 15 percentage points higher than GPT-5.6 Terra at about a third of the cost. It matches GPT-6 Astra at a tenth of Astra's per-token price. At max effort, OpenAI says Terra's cost per task is about nine times lower than Claude Opus 5.5's. On OpenAI's internal benchmark of error-inducing prompts, the error rate at low effort fell from 53.6% to 16.7%.

Independent testing by Artificial Analysis found that the new model is both more capable and much cheaper to run than GPT-5.6 Terra. At max effort it scores 6 points higher on the Artificial Analysis Intelligence Index, and its cost per task is $0.40, compared with $1.40. At lower effort settings the gain is larger, between 7 and 13 points. Artificial Analysis says the low and medium settings are Pareto-optimal for intelligence versus cost. At max effort, Terra scores 4 points below GPT-6.1 Sol and 10 points above GPT-6 Luna.

On the new Coding Agent Index v2.0, Terra scores 48. That puts it below Sol (54) and above Luna (31). Its xhigh and low effort settings sit on the cost-efficiency frontier.

The largest change Artificial Analysis measured is in hallucination. On AA-Omniscience, its private knowledge benchmark, Terra guesses 38% of the time when it does not know an answer, down from 88% for GPT-5.6 Terra. Its accuracy also rose by 2 points.

In Artificial Analysis's testing, Terra gained about 20 points on Terminal-Bench 4.0 and now rivals Sol at max effort on that benchmark. OpenAI reports a max-effort score of 50.3% there, against a published 21.5% for GPT-5.6 Terra. Gains on most other evaluations were smaller, such as 5 points on GDP.pdf.