GPT-6.1 Terra

Terra rolled out to Plus, Pro, Enterprise, and Edu users in ChatGPT Work and Codex, and developers can call it as gpt-6.1-terra. The context window is 1.05M tokens, with text and image input and six reasoning settings from none through max. On Agents' Last Exam at medium reasoning effort, Terra scores 15 percentage points above GPT-5.6 Terra at about one-third the cost, while matching GPT-6 Astra at one-tenth the price per token. Terra sits in the middle of OpenAI's model family. At the bottom is Luna, which is fast and inexpensive for simple queries. At the top are Sol and Astra, which are expensive but reliable even for complex tasks. GPT-6.1 Sol was released on September 29, 2026, during OpenAI's DevDay event. OpenAI also gave Terra a Critical cybersecurity rating under its Preparedness Framework, placing its security risks alongside those of GPT-6 Astra. On a harder exploit test, Terra achieved arbitrary code execution in 24.5% of cases, compared with 21.5% for GPT-6.1 Sol and 31.5% for GPT-6 Astra. For a team building an AI company, the change that matters most is cost per task for computer-use agents. At max effort, GPT-6.1 Terra gains 6 points in the Artificial Analysis Intelligence Index over GPT-5.6 Terra, at less than a third of the cost per task ($0.40 vs $1.40). Those are third-party numbers on a benchmark, so workflow tests on your own form-filling or screen tasks will tell you more. The Critical rating also means access controls and monitoring for agents built on this model deserve attention before they reach production.