OpenAI released GPT-6.1 Terra on Monday, October 26, as an upgrade to the middle model in OpenAI's lineup. Standard API prices are 50% below GPT-5.6 Terra. The reduction applies to the rates developers pay to integrate the model into applications. OpenAI says GPT-6.1 Terra performs computer tasks faster and more accurately.
Terra is designed for everyday office tasks, including interpreting screen contents, navigating applications and composing replies. The model's computer-use capability lets the AI view a screen and interact by clicking and typing. OpenAI said the rollout is underway for all Plus, Pro, Enterprise and Edu users through ChatGPT Work and Codex. Developers can access the model as gpt-6.1-terra.
ChatGPT Work is OpenAI's agent mode within ChatGPT, capable of operating across applications, browsing websites and completing website tasks. Codex is OpenAI's tool for coding. Free users have access to the lower-tier GPT-5.6 Luna.
OpenAI's internal test used prompts intended to induce model mistakes. OpenAI reported that factual errors at low reasoning effort declined from 53.6% for GPT-5.6 Terra to 16.7% for GPT-6.1 Terra. The reasoning-effort setting determines how long the model spends thinking. OpenAI reported that factual errors at maximum reasoning effort dropped by 18 points to 4.3%.
OpenAI reported that GPT-6.1 Terra completed 50.3% of Terminal-Bench 4.0 tasks at maximum reasoning effort, compared with 21.5% for GPT-5.6 Terra. Terminal-Bench 4.0 measures the proportion of tasks coding agents successfully complete through a computer's command-line terminal. OpenAI also says Terra completes tasks more than three times faster on average than GPT-6 Astra and GPT-6.1 Sol. OpenAI supplied all these performance figures, and the factual-error test was internal.
OpenAI first introduced Terra on June 26 within the GPT-5.6 family through a limited preview requested by the Donald Trump administration. OpenAI made Terra publicly available on July 9 as a model competitive with GPT-5.5 at approximately half the cost. OpenAI launched GPT-6 Sol and Luna in September, minutes after Anthropic released Claude Opus 5.5. That September launch did not include a new Terra model. OpenAI continues to recommend GPT-6.1 Sol or GPT-6 Astra for detailed research or complex coding.