OpenAI has launched the GPT-6.1 Terra model, which features API pricing at half the cost of the previous GPT-5.6 Terra version. Input tokens are priced at $1 per million, while output tokens cost $5 per million. Cached inputs are available at $0.05 per million tokens.
The model is designed for tasks such as drafting proposals, interpreting technical drawings, and filling out forms. It is currently accessible to Plus, Pro, Enterprise, and Edu users via ChatGPT Work, Codex, and the API. The company highlights its capabilities in visual reasoning and automated software control.
According to Artificial Analysis, the model achieves a higher score on the Intelligence Index compared to its predecessor while reducing costs per task. In the Coding Agent Index v2.0, Terra scored 48 points. The model also shows a reduction in hallucination rates, with guessing dropping to 38 percent from 88 percent.
On the Terminal-Bench 4.0 coding benchmark, the model solves 50.3 percent of tasks at maximum reasoning effort, compared to 21.5 percent for the previous version. While it shows significant efficiency, OpenAI recommends using GPT-6.1 Sol or GPT-6 Astra for the most demanding coding requirements.
GPT-6.1 Terra meets the company's internal cybersecurity threshold and includes safeguards inherited from GPT-6 Astra. The model demonstrates a 98.8 percent defense rate against indirect prompt injection attacks, an improvement over the 96.2 percent rate observed in GPT-5.6 Terra.
