OpenAI has launched GPT-6.1 Terra, a mid-tier model designed for workflows involving multiple tools, computer use and routine coding. OpenAI says it completes agent tasks more than three times faster on average than GPT-6 Astra and GPT-6.1 Sol.
From filling forms to reviewing code
Terra’s intended tasks include navigating applications and filling out forms. It is also designed to edit documents, review code and coordinate multiple tools within a workflow.
For more demanding tasks, OpenAI continues to recommend Sol or Astra. These include deep research, difficult coding, mathematical proofs and long chains of reasoning.
OpenAI says Terra matches Astra in a workplace software test
OpenAI reports a substantial improvement over GPT-5.6 Terra on Agents’ Last Exam, a benchmark of how models use workplace software. The company says the new Terra also matches GPT-6 Astra on that test.
Access through ChatGPT Work, Codex and the API
GPT-6.1 Terra is coming to ChatGPT Work and Codex, with access rolling out to users on Plus, Pro, Enterprise and Edu plans. Developers can call the model through the API using the identifier gpt-6.1-terra.
Standard API rates are half those of GPT-5.6 Terra: one million input tokens cost $1 rather than $2, while one million output tokens cost $6 rather than $12.
OpenAI also says Terra uses up allowances more slowly than Sol or Astra in ChatGPT Work and Codex.
