OpenAI released GPT-6.1 Terra on Monday, October 26, 2026 — a mid-tier model built for computer use, visual reasoning and everyday writing, priced at half of what GPT-6.1 Sol costs. Why it matters: Terra is OpenAI’s bet that the next wave of AI spending goes to agents doing routine computer work repeatedly, where cost per finished task matters more than peak intelligence.
October 26, 2026
What happened
Terra fills the one gap left in OpenAI’s current lineup. When the company moved Sol and Luna onto the GPT-6 generation in September, Terra stayed behind on GPT-5.6. GPT-6.1 Terra now replaces GPT-5.6 Terra and sits between Luna and Sol in a four-tier range, with GPT-6 Astra at the top. It is available now in ChatGPT Work and Codex for Plus, Pro, Enterprise and Edu users, and to developers through the API as gpt-6.1-terra. Vercel added it to its AI Gateway the same day.
The details
OpenAI pitches Terra at work that combines looking at a screen with acting on it: drafting proposals, reading technical drawings, filling out forms, editing video, and entering invoice details into accounting software. Standard API pricing is $1 per million input tokens, $0.05 per million cached input tokens and $5 per million output tokens — 50% below both GPT-5.6 Terra and GPT-6.1 Sol. The model accepts text and images, returns text, and has a context window of more than 1 million tokens.
On OpenAI’s own reported benchmarks, Terra scored 15 percentage points above GPT-5.6 Terra on Agents’ Last Exam v1 at medium reasoning effort, at about one-third the cost, and matched GPT-6 Astra on that test at one-tenth of Astra’s per-token price. On Terminal-Bench 4.0 it solved 50.3% of tasks at maximum reasoning effort, up from GPT-5.6 Terra’s published 21.5%. On the company’s internal set of error-inducing prompts, the error rate at low reasoning effort fell from 53.6% to 16.7%, and to 4.3% at maximum effort. OpenAI also says Terra finished benchmark computer tasks more than three times faster than Astra and Sol, and that it meets the Critical threshold for cybersecurity under the company’s Preparedness Framework. Those are vendor-reported figures on vendor-chosen tests, so treat them as a reason to test Terra on your own work, not as settled rankings.
Why it matters
Everything AI, tested means asking what a model finishes reliably for the money — and that is exactly the contest Terra enters. A cheaper model that drives software competently changes the math for high-volume office automation: the same ChatGPT Work or Codex usage allowance covers more tasks with Terra than with Sol or Astra, according to OpenAI. OpenAI still recommends Sol and Astra for the most complex coding, which tells you where Terra’s ceiling is meant to be. If its real-world completion rate holds up, expect routine document and desktop work to migrate down to this price tier quickly.
Frequently asked questions
How much does GPT-6.1 Terra cost?
Standard API rates are $1 per million input tokens, $0.05 per million cached input tokens and $5 per million output tokens — half the price of GPT-6.1 Sol.
Where can I use GPT-6.1 Terra?
In ChatGPT Work and Codex on Plus, Pro, Enterprise and Edu plans, and through the OpenAI API as gpt-6.1-terra. It is not offered in ordinary ChatGPT chat.
What is GPT-6.1 Terra best for?
OpenAI positions it for visual reasoning and computer-use work — forms, invoices, technical drawings, proposals and video edits — rather than the hardest reasoning or coding jobs, which it still steers to Sol and Astra.
Sources: OpenAI announcement (via AI Affairs and AIstify reporting), Seeking Alpha, Vercel AI Gateway changelog