OpenAI releases GPT-6.1 Terra, with API pricing at half that of GPT-5.6 Terra
On October 26, 2026, OpenAI released and made available GPT-6.1 Terra, strengthening visual understanding, computer operation, and writing capabilities. The standard API price is half that of GPT-5.6 Terra, namely $1 per million tokens for input, $0.05 for cached input, and $5 for output. The model has a context window of 1.05 million tokens, a maximum input of 922,000 tokens, and a maximum output of 128,000 tokens. Reasoning effort can be selected from none to max, with medium as the default. On Agents' Last Exam, the medium setting scores 15 points higher than GPT-5.6 Terra while costing about one-third as much; under the max setting, the average speed for completing tasks is more than 3 times that of GPT-6 Astra and GPT-6.1 Sol. The model is already available to Plus, Pro, Enterprise, and Edu users of ChatGPT Work and Codex, and the API model ID is gpt-6.1-terra.
Why read this
SelectedAI score 90
The original article lists the new model's benchmark performance on visual understanding and computer operation as well as changes in API pricing, which can be used to judge cost trade-offs for everyday automation tasks.
Topics
Body · AI translation
On October 26, 2026, OpenAI announced and began offering "GPT-6.1 Terra," a new version of its AI model focused on balancing both performance and cost.
Introducing GPT-6.1 Terra | OpenAI
GPT-6.1 Terra: delivering fast and capable computer use at a lower price.
GPT-6.1 Terra understands what’s on screen, takes action across apps, writes naturally, and is a more capable model for your whole workday with the efficiency to take on more of it. https://t.co/8JTN1yonmV— OpenAI (@OpenAI) October 26, 2026
GPT-6.1 Terra is a major update to the previous version, GPT-5.6 Terra, and is now capable of detailed visual reasoning, precise computer operation, and clearer, more natural writing. It also greatly improves the balance of speed, performance, and cost, with the standard API price set at half that of GPT-5.6 Terra. Furthermore, when used with ChatGPT Work and Codex, the tasks that can be covered under the same usage limits greatly exceed those of GPT-5.6 Terra as well as the higher-tier models GPT-6.1 Sol and GPT-6 Astra.
The context window is 1.05 million tokens, with a maximum input of 922,000 tokens and a maximum output of 128,000 tokens. Reasoning effort (the amount of processing devoted to reasoning) can be specified as none, low, medium, high, xhigh, or max, with medium as the default.
Improved ability to understand images and connect them to work
According to the company, GPT-6.1 Terra is suited to automating large volumes of work that involve visual understanding, such as reading large numbers of invoices and entering the details into accounting software. When performance and cost were compared on "Agents' Last Exam," a benchmark that measures performance on computer tasks in the workplace, GPT-6.1 Terra with reasoning effort set to medium recorded a score 15 points higher than GPT-5.6 Terra at about one-third the cost, including on tasks such as building 3D models from architectural drawings. Additionally, when reasoning effort was set to max, it completed tasks more than three times faster than the average speed of GPT-6 Astra and GPT-6.1 Sol.
Furthermore, on "MMMU-Pro," which measures the ability to reason based on the content of images, GPT-6.1 Terra outperformed GPT-5.6 Terra under every reasoning setting evaluated.
GPT-6.1 Terra substantially outperforms GPT-5.6 Terra on Agents’ Last Exam, a test of how well agents use software at work.
It even matches GPT-6 Astra at one-tenth the price per token and finishes tasks over 3× faster on average than both Astra and GPT-6.1 Sol. pic.twitter.com/pzAmnvjBUi— OpenAI (@OpenAI) October 26, 2026
Coding ability and factual accuracy also improved
GPT-6.1 Terra also has improved coding performance. On "Terminal-Bench 4.0," which measures the ability to solve development tasks on the command line, the solve rate of GPT-6.1 Terra with reasoning effort set to max greatly exceeded that of GPT-5.6 Terra. On the other hand, GPT-6.1 Sol and GPT-6 Astra remain better suited to more complex coding work.
Another major improvement is factual accuracy. When factual accuracy was evaluated using conversations in which users reported factual errors in the responses of previous models, the error rate of GPT-6.1 Terra with reasoning effort set to low was about one-third that of GPT-5.6 Terra. See the system card for detailed evaluation methods and results.
Security measures will be implemented continuously
GPT-6.1 Terra retains the safeguards developed for GPT-6 Astra. According to OpenAI's evaluation, GPT-6.1 Terra tends to stay in scope more readily than GPT-5.6 Terra and to honestly disclose limitations and potential failures to users. In prompt injection tests as well, GPT-6.1 Terra reportedly showed strong resistance to malicious instructions intended to steer it away from the user's task. This is also reported in detail in the system card.
GPT-6.1 Terra is available to all Plus, Pro, Enterprise, and Edu users of ChatGPT Work and Codex. It is also available via the API (model ID: gpt-6.1-terra). Standard API pricing is half that of GPT-5.6 Terra: $1 per million tokens for input, $0.05 for cached input, and $5 for output. When input exceeds 272,000 tokens, input and cached pricing are doubled and output pricing is 1.5 times higher for the entire request, not just the excess portion. Fast mode pricing is twice the standard rate, and some tools incur separate call charges. See the API pricing and terms of use for details.