OpenAI has introduced GPT-6.1 Terra for visual reasoning, writing and computer tasks, with standard API prices 50% below those of GPT-5.6 Terra and GPT-6.1 Sol. The model is available today through the API and to Plus, Pro, Enterprise and Edu subscribers using ChatGPT Work and Codex. OpenAI says those subscribers can complete more tasks within their existing usage allowance than with GPT-5.6 Terra.
Developers pay $1 for each million input tokens, $0.05 for each million cached input tokens and $5 for each million output tokens. The API identifier is gpt-6.1-terra, and OpenAI describes invoice processing and entering data into accounting software as potential uses. Despite the coding gains, OpenAI still recommends GPT-6.1 Sol and GPT-6 Astra for the hardest programming work.
Read more: Chinese users praise OpenAI's GPT-5.6 for efficiency despite higher prices than local rivals
On Agents’ Last Exam, which tests professional computer tasks, OpenAI reports a 15 percentage point improvement over GPT-5.6 Terra at medium reasoning effort for roughly a third of the cost. At that setting, the company says GPT-6.1 Terra matches GPT-6 Astra while charging one-tenth as much per token. OpenAI also reports task completion more than three times as fast as GPT-6 Astra and GPT-6.1 Sol.
OpenAI’s internal tests using prompts designed to provoke errors recorded a 16.7% error rate at low reasoning effort, compared with 53.6% for GPT-5.6 Terra. At maximum effort, the new model’s error rate was 4.3%, an improvement of 18 percentage points. For command-line coding tasks on Terminal-Bench 4.0, maximum-effort performance reached 50.3%, compared with the predecessor’s published 21.5%.
The company says GPT-6.1 Terra reaches its Critical cybersecurity threshold under the Preparedness Framework and retains safeguards developed for GPT-6 Astra. OpenAI’s evaluations also found that it more often respected task boundaries and disclosed limitations than GPT-5.6 Terra. The company reports resistance to prompt injections intended to redirect the model away from a user’s task.
The release follows OpenAI’s October 7 introduction of GPT-6 with Intelligent UI in ChatGPT, which added interactive responses such as diagrams, calculators and forms. That earlier update applied to Chat, leaving the models serving Work and Codex unchanged.
