OpenAI released GPT-6.1 Terra on October 26 for ChatGPT Work, Codex and its API, a mid-tier model built for visual reasoning and computer use priced at $1 per million input tokens and $5 per million output tokens, half the rates of GPT-6.1 Sol.
OpenAI released GPT-6.1 Terra on October 26, 2026, a mid-tier model for ChatGPT Work and Codex and for developers through the API under the identifier gpt-6.1-terra. The company positions it as the successor to GPT-5.6 Terra, built for visual reasoning, precise computer use and clearer writing, with everyday office work as its target: drafting proposals, reading technical drawings, filling in forms, editing video, and logging incoming invoices into accounting software.
The price is the headline. Standard API rates are $1 per million input tokens, $0.05 per million cached input tokens and $5 per million output tokens, half the rates of GPT-6.1 Sol and half what GPT-5.6 Terra cost. Prompts above 272,000 input tokens attract higher rates across the full request. The context window is 1,050,000 tokens with a maximum output of 128,000 tokens, and the knowledge cutoff is June 1, 2026.
OpenAI's own benchmarks show large gains over the previous Terra. On Terminal-Bench 4.0, Terra solves 50.3% of tasks at maximum effort against 21.5% for GPT-5.6 Terra, and on its internal difficult-prompts evaluation the factual-error rate fell from 53.6% to 16.7% at low reasoning effort and to 4.3% at maximum effort. On Agents' Last Exam at maximum effort, OpenAI's published chart puts Terra at 60.0% and $1.01 per task, against 63.2% and $8.82 for Anthropic's Claude Opus 5.5.
Those benchmark figures are vendor-reported and have not been independently verified, with one exception: Artificial Analysis, an independent benchmarking firm, published its own evaluation on launch day, giving Terra an Intelligence Index of 48.1 at $0.40 per task, against $1.40 per task for GPT-5.6 Terra, and a Coding Agent Index of 48 against 54 for Sol. OpenAI itself says Sol and Astra remain better choices for the most complex coding tasks.
Availability covers ChatGPT Work and Codex for Plus, Pro, Business, Enterprise and Edu plans, with rollout varying by plan, region and workspace settings, plus API access and distribution through Amazon Bedrock and GitHub Copilot. The model accepts text and images and produces text; native audio and video are not supported, and tool calling requires the Responses API.
The stakes are about the cost of letting models do things rather than just answer questions. Terra is not a frontier model on raw capability, but at roughly a ninth of the per-task cost OpenAI claims against its leading rival, it changes the arithmetic for businesses running agents at volume. The next test is independent: whether real-world deployments match the launch-day benchmarks.