OpenAI cuts costs and speeds up desktop automation with its midtier GPT-6.1 Terra model

OpenAI has debuted GPT-6.1 Terra, a midtier artificial intelligence model designed to automate everyday computer tasks at half the operating cost of its predecessor. The update is available immediately through the API at $1 per million input tokens and $5 per million output tokens, as well as in ChatGPT Work and Codex for Plus, Pro, Enterprise, and Edu subscribers. [1][3][4]

The company built Terra to handle desktop chores such as editing video, filling out forms, and interpreting technical drawings. OpenAI says the system finishes tasks more than three times faster than higher-end models like GPT-6 Astra and GPT-6.1 Sol. It also posts significant benchmark gains, cutting internal error rates at maximum reasoning effort to 4.3% and scoring 50.3% on Terminal-Bench 4.0, more than doubling the programming mark set by GPT-5.6 Terra. [3][4]

Terra sits between the lightweight Luna and the more capable Sol and Astra tiers in OpenAI's lineup. While OpenAI matches Terra's performance against Astra at a fraction of the token cost on certain agent evaluations, it still advises developers to turn to Sol or Astra when tackling the most intricate coding challenges. [3][4]