OpenAI has introduced GPT-6.1 Terra, a new AI model designed to handle computer-based tasks, visual reasoning, coding and everyday professional work more quickly and efficiently. Announced on October 26, 2026, the model succeeds GPT-5.6 Terra and focuses on helping AI systems understand what is displayed on a screen and take actions across digital applications.

The release also brings lower API pricing and improvements in factual accuracy, according to OpenAI. The company positions GPT-6.1 Terra as a practical option for workflows such as processing invoices, filling out forms, interpreting technical drawings, drafting proposals and editing videos.
What Is GPT-6.1 Terra?
GPT-6.1 Terra is part of OpenAI’s GPT-6 model family. It is designed to balance capability, speed and operating cost, particularly for tasks that involve visual information and interaction with software.
Rather than focusing only on generating text, the model can interpret screen content and use supported computer tools to carry out multistep workflows. These capabilities are intended to help developers build agents that can move between applications and complete structured tasks.
OpenAI describes Terra as a fast, efficient model with improved visual understanding, computer-use capabilities and more natural writing. It sits within the company’s broader model lineup alongside options designed for different levels of reasoning capability and cost.
Faster Computer Use Across Everyday Applications
One of the main goals of GPT-6.1 Terra is to improve how AI agents interact with desktop software and other digital interfaces.
Computer-use workflows can include reading information on a screen, clicking interface elements, entering data, navigating between applications and checking the results of an action. These capabilities could support repetitive office work that traditionally requires people to move information manually between systems.
Potential use cases include:
- Reading incoming invoices and entering details into accounting software.
- Completing forms and transferring information between applications.
- Interpreting technical drawings and other visual documents.
- Drafting and revising proposals, reports and business communications.
- Reviewing video footage and creating time-coded notes.
- Supporting coding workflows and debugging software.
These are intended use cases, not a guarantee that every task will succeed without supervision. Reliability will depend on the application, the quality of the available information and how the workflow is configured.
OpenAI Reports Improvements on Computer-Use Benchmarks
OpenAI evaluated GPT-6.1 Terra on several benchmarks covering professional tasks, visual reasoning and software engineering.
On Agents’ Last Exam, a benchmark featuring complex professional tasks, OpenAI reported that Terra scored 15 percentage points higher than GPT-5.6 Terra at medium reasoning effort, at approximately one-third of the cost. The company also said Terra completed tasks more than three times faster than GPT-6 Astra and GPT-6.1 Sol in its reported comparison.
The company also reported improvements on OSWorld 2.1, an evaluation of agents performing multistep tasks in desktop applications, and MMMU-Pro, a benchmark for visual reasoning.
These results offer an indication of how OpenAI says the model performs under specific test conditions. They should not be interpreted as a guarantee of the same speed or accuracy across all real-world applications.
Read More:- Anthropic Reveals AI Security Incidents in IPO Filing
Lower API Pricing Could Make AI Automation More Accessible
Pricing is another important part of the GPT-6.1 Terra release.
OpenAI lists standard API prices of $1 per million input tokens, $0.05 per million cached input tokens and $5 per million output tokens. The company says its standard pricing is 50% lower than GPT-5.6 Terra and GPT-6.1 Sol.
Lower token costs could make it more economical to run high-volume workflows, especially when an agent processes many documents or completes repetitive tasks throughout the day. However, the total cost of a computer-use workflow may also include tool calls and other usage charges, so token pricing alone does not determine the final bill.
For businesses building AI agents, the combination of computer interaction and lower per-token pricing may make it easier to evaluate automation for tasks that were previously too expensive to run at scale.
Improvements in Factual Accuracy and Coding
OpenAI also reports a substantial reduction in factual errors on its internal benchmark of prompts designed to expose mistakes.
At low reasoning effort, the company says the error rate fell from 53.6% for GPT-5.6 Terra to 16.7% for GPT-6.1 Terra. At maximum reasoning effort, OpenAI reported an error rate of 4.3%.
For coding, OpenAI reported that GPT-6.1 Terra solved 50.3% of tasks on Terminal-Bench 4.0 at maximum reasoning effort, compared with a published score of 21.5% for GPT-5.6 Terra.
These figures come from OpenAI’s own evaluations and reflect particular benchmarks and settings. They do not mean the model is free of hallucinations or that it will outperform every alternative in every coding task.
OpenAI notes that its more capable models remain better suited to some of the most complex reasoning and software-engineering work.
Safety and Availability
OpenAI says GPT-6.1 Terra retains protections developed for GPT-6 Astra and meets the “Critical” threshold for cybersecurity under the company’s Preparedness Framework. It also reports improvements in staying within task boundaries and resisting prompt-injection attempts designed to redirect an AI agent.
Such safeguards are particularly relevant for computer-use systems, because an agent may encounter untrusted instructions inside web pages, documents or application interfaces. Users and developers should still apply appropriate permissions, monitoring and human review for consequential actions.
According to OpenAI, GPT-6.1 Terra is available to Plus, Pro, Enterprise and Edu users through ChatGPT Work and Codex. Developers can access it through the API using the model identifier gpt-6.1-terra.
The model accepts text and image inputs, while its published API documentation lists audio and video as unsupported input modalities for the model itself.
What GPT-6.1 Terra Means for the Future of AI Work
The launch reflects a broader shift from AI tools that mainly answer questions toward agents that can interact with software and complete multistep tasks.
If models become more reliable at understanding screens, handling instructions and using applications, organizations could automate more routine administrative and technical work. Lower operating costs could further encourage businesses to test these systems in document processing, coding support and other repetitive workflows.
However, successful deployment will require more than model performance. Companies will need to consider data access, permissions, error handling, privacy and the consequences of an agent taking an incorrect action.
GPT-6.1 Terra is OpenAI’s latest attempt to combine computer-use capability with speed and cost efficiency. Its practical impact will depend on how well those capabilities translate from benchmark results into dependable work in real applications.
Frequently Asked Questions
What is GPT-6.1 Terra?
GPT-6.1 Terra is an OpenAI model designed for visual reasoning, computer use, coding and professional tasks. It succeeds GPT-5.6 Terra.
What can GPT-6.1 Terra do?
It can interpret visual information and, through supported tools, interact with computer interfaces. Potential tasks include filling out forms, processing invoices, interpreting technical drawings and assisting with coding.
Is GPT-6.1 Terra cheaper than GPT-5.6 Terra?
OpenAI says standard API pricing is 50% lower than GPT-5.6 Terra. Published prices are $1 per million input tokens and $5 per million output tokens, with cached input priced at $0.05 per million tokens.
Who can use GPT-6.1 Terra?
OpenAI says it is available to Plus, Pro, Enterprise and Edu users through ChatGPT Work and Codex. Developers can use it through the API.
Does GPT-6.1 Terra eliminate AI errors?
No. OpenAI reports lower error rates on its internal evaluations, but benchmark improvements do not eliminate hallucinations or guarantee accuracy in every real-world task.




