Recently, OpenAI officially launched the all-new generation AI model GPT-6.1Terra. The model focuses its core optimization on autonomous computer operation, visual understanding, and complex programming capabilities, fully empowering users to efficiently complete everyday high-frequency tasks such as writing proposals, interpreting technical drawings, filling in spreadsheets, and editing videos.
In terms of operational efficiency and core performance, GPT-6.1Terra has achieved a significant leap. Data shows that the model completes computer automation tasks more than 3 times faster than the previous generation. At the same time, its error rate has dropped substantially: the error rate at the low reasoning tier fell sharply from the previous 53.6% to 16.7%, while the error rate at the highest reasoning tier was pushed down to 4.3%, a full 18 percentage points lower than the previous generation. In highly challenging programming tests, the score at its highest reasoning tier also jumped from 21.5% in the previous generation to 50.3%.
Alongside the strong performance upgrade, its invocation costs have been significantly reduced. According to the officially announced API pricing, GPT-6.1Terra's API price has dropped by 50% compared to before, with specific rates of $1 per million input tokens, $0.05 for cached input tokens, and $5 for output tokens.
Currently, this new model has been fully rolled out to all user tiers of ChatGPT Work and Codex, including Plus, Pro, Enterprise, and Edu, while developers can also conveniently integrate it via the API.