OpenAI unveils new model with particularly efficient and fast performance

OpenAI has unveiled the GPT-6.1 Terra model, a new intermediate model for routine office work, half the price of the previous generation and three times faster. The hallucination error rate dropped from 53.6% to 16.7% at low effort. The model is available to ChatGPT Plus, Pro, Enterprise, and Edu subscribers, and to developers via the API.

OpenAI has unveiled the GPT-6.1 Terra model, a new intermediate model in the sixth-generation family, designed for routine office work and continuous computer control. The model is half the price of the previous generation – one dollar per million input tokens and five dollars per million output tokens – and is three times faster than the flagship Astra and Sol models. In a significant achievement, the hallucination error rate dropped from 53.6% in the previous generation to just 16.7% at low computational effort, and up to 4.3% at maximum effort. In the Agents' Last Exam, the model scored 15 percentage points higher than its predecessor at a third of the cost. The model is available to ChatGPT Plus, Pro, Enterprise, and Edu subscribers, and to developers via the API.