OpenAI has launched GPT-6.1 Terra, a new model designed for efficient computer operation and visual reasoning. The model is now available to Plus, Pro, Enterprise, and Edu users via ChatGPT Work and Codex, and is accessible to developers through the OpenAI API under the identifier gpt-6.1-terra.
The model features significant cost reductions compared to previous versions. Its standard API pricing is 50% lower than GPT-5.6 Terra and GPT-6.1 Sol. In terms of performance, GPT-6.1 Terra matches GPT-6 Astra in computer operation capabilities while costing only one-tenth per token. It also operates three times faster than GPT-6 Astra and GPT-6.1 Sol.
In benchmark testing, GPT-6.1 Terra demonstrated improved accuracy. On an internal error-prone prompt test, the error rate dropped from 53.6% in GPT-5.6 Terra to 16.7% at medium reasoning intensity. In Terminal-Bench 4.0, the model successfully completed 50.3% of programming tasks, compared to 21.5% for its predecessor.
The model includes enhanced safety features, reaching a 'Critical' capability threshold in cybersecurity according to OpenAI's Preparedness Framework. It also shows improved resistance to prompt injection attacks and better adherence to user-specified operational boundaries compared to the GPT-5.6 Terra version.
