OpenAI introduced GPT-6.1 Terra on Tuesday, describing it as a fast, efficient model that combines detailed visual reasoning and precise computer use with clearer, more natural writing. The model is available starting October 27, 2026, in ChatGPT Work and Codex for Plus, Pro, Enterprise and Edu users, and to developers through the OpenAI API under the identifier gpt-6.1-terra.
The pitch is everyday work done on a computer, not just conversation. OpenAI lists drafting proposals, interpreting technical drawings, filling out forms and editing videos among the intended uses. For developers, the company points to high-volume automation, such as reading incoming invoices and entering their details into accounting software, as the kind of task where a cheaper, faster model changes what is practical to automate.
On computer-use tests reported by OpenAI, the gains over GPT-5.6 Terra are large. On Agents' Last Exam, a benchmark of professional tasks that includes converting architectural drawings into 3D models and reviewing an animated film shot by shot, Terra scored 15 percentage points higher than its predecessor at medium reasoning effort, at about one-third of the cost. OpenAI says its performance was also on par with GPT-6 Astra at one-tenth of the price per token, and that at maximum effort it completed tasks more than three times faster than GPT-6 Astra and GPT-6.1 Sol.
Accuracy is the other headline claim. On an internal benchmark of error-inducing prompts, OpenAI reports Terra's error rate fell to 16.7 percent at low reasoning effort, compared with 53.6 percent for GPT-5.6 Terra, and to 4.3 percent at maximum effort. On Terminal-Bench 4.0, which tests coding agents on command-line tasks, Terra solved 50.3 percent of tasks at maximum effort, up from 21.5 percent for GPT-5.6 Terra. OpenAI is careful to note that GPT-6.1 Sol and GPT-6 Astra remain the better choices for the most complex coding work.
On safety, OpenAI says Terra meets the Critical threshold for cybersecurity under its Preparedness Framework and retains the safeguards developed for GPT-6 Astra. In its evaluations, the company says the model is more likely than its predecessor to stay within the scope of a task and to disclose limitations and blockers honestly, and that it showed resistance in prompt-injection tests to malicious instructions meant to divert it from the user's task. Those are OpenAI's own evaluations, and independent testing will matter as developers put the model to work.
The pricing is what makes the launch a broader industry story. Standard API rates are 1 dollar per million input tokens, 0.05 dollars per million cached input tokens and 5 dollars per million output tokens — rates OpenAI says are 50 percent lower than those of GPT-5.6 Terra and GPT-6.1 Sol. In ChatGPT Work and Codex, the company says the same usage allowance covers more tasks with Terra than with the earlier model. If the benchmark claims hold up in real use, Terra pressures rivals on cost per completed task, not just headline capability — and gives teams a reason to route routine computer work to a cheaper model while reserving the largest models for the hardest problems.