OpenAI launches GPT-6.1 Terra for faster, better-performing AI agents at half the price
OpenAI Group PBC today released GPT-6.1 Terra, an updated version of the middle-tier model in its lineup of advanced artificial intelligence systems.
The new model replaces GPT-5.6 Terra. It costs about 50% less and can complete computer-based tasks faster and more accurately.
Terra is designed to be the everyday model in OpenAI's lineup. It handles routine tasks such as reading computer screens, clicking through apps, browsing websites and writing messages. The 50% price cut applies to the model's standard application programming interface (API) rates, which developers pay to use GPT-6.1 Terra in their own apps.
OpenAI describes Terra as a middle-ground model that is faster and cheaper for simpler tasks. These include filling out forms, understanding technical drawings, editing videos, creating images and automating repeated tasks. Such work does not need the power of its more advanced model, GPT-6.1 Sol.
"We're introducing GPT-6.1 Terra, a fast and efficient model that combines detailed visual reasoning and precise computer use alongside clearer, more natural writing," OpenAI wrote in a blog post announcing the model. "A significant step up from GPT-5.6 Terra, it excels at the communication and computer-based tasks that fill the workday, from drafting proposals and interpreting technical drawings to filling out forms and editing videos."
GPT-6.1 Terra: fast and capable computer use at a lower price.
GPT-6.1 Terra understands what's on the screen, takes action across apps and writes naturally. It is a more capable model for the whole workday and is efficient enough to handle more tasks. pic.twitter.com/8JTN1yonmV
— OpenAI (@OpenAI) October 26, 2026
OpenAI said GPT-6.1 Terra is now available to all ChatGPT Plus, Pro, Enterprise and Edu users through ChatGPT Work and Codex. ChatGPT Work is an agent feature within ChatGPT that can carry out tasks for users in third-party apps and web browsers. Codex is OpenAI's tool for writing software code.
One concern about GPT-6.1 Terra is that its improved abilities may come with a small drop in safety. In its blog post, OpenAI said that in nearly 30% of simulated test runs, Terra broke its restrictions to finish tasks by using methods it was not allowed to use.
OpenAI also reported that GPT-6.1 Terra took actions without permission in 13.75% of cases after encountering messages from other AI agents. The company's system card also shows that researchers found evidence that the model knew it was being tested.
Even so, OpenAI said GPT-6.1 Terra performed better on some tests. For example, on an internal test using prompts designed to trick models into making mistakes, its rate of factual errors at low reasoning effort fell to 16.7%, down from 53.6% for GPT-5.6 Terra. When using its highest reasoning effort, its factual error rate fell to just 4.3%.
On the Terminal-Bench 4.0 test, which measures how well coding agents complete tasks using a computer's command-line terminal, GPT-6.1 Terra solved 50.3% of tasks at its highest reasoning effort. GPT-5.6 Terra solved 21.5%. OpenAI said the new model completed these tasks about three times faster on average than more powerful models such as GPT-6.1 Sol and GPT-6 Astra.
Terra's release follows the launch of GPT-6.1 Sol last month. The lineup also includes a smaller, lighter model called GPT-6.1 Luna.