OpenAI Launches GPT-6.1 Terra, a Faster Computer Use Model at Half the Price
OpenAI has launched GPT-6.1 Terra, a fast and efficient model built for the computer based work that fills a typical workday. Announced on October 26, 2026, the model is positioned as a significant step up from GPT-5.6 Terra, with detailed visual reasoning, precise computer use, and clearer, more natural writing.
Terra is aimed at practical tasks rather than open ended chat. OpenAI describes its strengths as drafting proposals, interpreting technical drawings, filling out forms, and editing videos. In other words, it is built to look at what is on screen and then act on it, moving between applications to complete multi step work.
Background: where Terra fits in the GPT-6 family
Terra sits in the middle of OpenAI’s current model range. GPT-6 Astra is the most capable frontier model, GPT-6.1 Sol sits above Terra for more complex work, and GPT-6 Luna is the smaller entry level tier. Terra is the fast mid tier option, designed for high volume work where speed and cost per task matter as much as raw capability.
Key details: price, availability, and performance
Price is central to the launch. Standard API pricing for GPT-6.1 Terra is 1 dollar per million input tokens, 0.05 dollars per million cached input tokens, and 5 dollars per million output tokens. OpenAI says those standard rates are 50 percent lower than the rates for GPT-5.6 Terra and GPT-6.1 Sol.
Terra is available to Plus, Pro, Enterprise, and Edu users in ChatGPT Work and Codex. Developers can access it through the OpenAI API under the identifier gpt-6.1-terra. In ChatGPT Work and Codex, OpenAI says the same usage allowance covers more tasks with Terra than with GPT-5.6 Terra, and substantially more than with GPT-6.1 Sol or GPT-6 Astra.
On computer use, OpenAI measured Terra on Agents Last Exam, a benchmark of professional tasks that includes turning architectural drawings into a 3D model and reviewing an animated film shot by shot to create a time coded log for a restoration team. At medium reasoning effort, OpenAI reports that Terra scores 15 percentage points higher than GPT-5.6 Terra at about one third the cost. It also reports that Terra performs on par with GPT-6 Astra at only one tenth of Astra’s price per token, and completes tasks more than three times faster than both GPT-6 Astra and GPT-6.1 Sol.
On MMMU-Pro, a benchmark of visual reasoning, OpenAI says Terra outperforms GPT-5.6 Terra at every reasoning setting and nearly matches GPT-6 Astra and GPT-6.1 Sol at much lower cost. On OSWorld 2.1, which tests how well agents use desktop applications, Terra surpasses the best score from GPT-5.6 Terra at a fraction of the cost per task at medium reasoning effort.
Factual accuracy and coding are the other headline improvements. On OpenAI’s internal benchmark of prompts designed to induce errors, the error rate at low reasoning effort fell from 53.6 percent with GPT-5.6 Terra to 16.7 percent with GPT-6.1 Terra. At maximum effort, the error rate fell by 18 percentage points to 4.3 percent, nearly matching the rates for GPT-6 Astra and GPT-6.1 Sol.
On Terminal-Bench 4.0, which tests how well coding agents solve problems from the command line, Terra solves 50.3 percent of tasks at maximum reasoning effort, more than double the published 21.5 percent score for GPT-5.6 Terra. OpenAI is clear about the limit here: for the most complex coding tasks, GPT-6.1 Sol and GPT-6 Astra remain the better choice.
Safety is also part of the announcement. OpenAI says Terra meets the Critical threshold in cybersecurity under its Preparedness Framework and maintains the protections developed for GPT-6 Astra. The company also reports alignment improvements, including a greater likelihood of staying within scope and honestly disclosing limitations, and strong resistance in prompt injection tests.
Independent checks add useful context
Early independent testing, as reported by WinBuzzer, broadly supports the direction of OpenAI’s claims while showing where Terra still trails larger models. Independent evaluator Artificial Analysis measured an approximately 20 percentage point gain over GPT-5.6 Terra on Terminal-Bench 4.0, and found that at maximum effort Terra gained six points on its general Intelligence Index while the weighted average cost per evaluated task fell from 1.40 dollars to 0.40 dollars.
Other tests show a more mixed picture. In a 43 question analytics exam from Plotly, Terra passed 40 questions, compared with 43 for GPT-6.1 Sol and 38 for the smaller GPT-6 Luna. In a 101 case accounting test from DualEntry, Terra averaged 78.2 percent accuracy, with weaker results on payables, receivables, and month end close cases. Those results are useful reminders that speed and low cost still need human review for recurring operational work.
Why it matters
Terra matters because much of the value in AI tools now comes from routine, high volume work, not from a single impressive demonstration. OpenAI gives the example of reading incoming invoices and entering their details into accounting software. That kind of task is repeated thousands of times, so a model that is fast enough, accurate enough, and cheap enough can change whether automation is worth building at all.
At half the standard API price of GPT-6.1 Sol, Terra lowers the cost of building agents that read screens and fill forms. Teams that need the strongest performance on the hardest coding problems should still look to Sol or Astra.
