OpenAI has released the GPT-6.1 Terra model, which is designed to perform complex computer operations and desktop software automation at a lower cost. The model is specifically optimized for tasks involving visual reasoning, such as processing invoices or interpreting technical drawings.
In the Agents’ Last Exam benchmark, Terra achieved a 15% higher score than the GPT-5.6 Terra at medium reasoning intensity, while reducing task costs by approximately two-thirds. Compared to GPT-6 Astra, Terra offers similar performance at one-tenth the cost per token.
The new model demonstrates significant improvements in accuracy and speed. At low reasoning intensity, the error rate dropped to 16.7% from 53.6% in the previous generation. At maximum intensity, the error rate is 4.3%, and task execution speed is more than three times faster than GPT-6 Astra and GPT-6.1 Sol.
Programming capabilities have also been enhanced, with a 50.3% success rate in Terminal-Bench 4.0, compared to 21.5% for GPT-5.6 Terra. However, OpenAI recommends using GPT-6.1 Sol or GPT-6 Astra for the most demanding programming tasks.
GPT-6.1 Terra is available to Plus, Pro, Enterprise, and Edu users via ChatGPT Work, Codex, and the OpenAI API. API pricing is set at $1 per million input tokens, $0.05 per million cached input tokens, and $5 per million output tokens, representing a 50% price reduction compared to previous models.
