OpenAI’s GPT-6.1 Terra Arrives: AI Takes Control of Your PC

Ender Öztürk

October 26, 2026

OpenAI has officially introduced its latest artificial intelligence model, GPT-6.1 Terra, designed to revolutionize computer interaction and visual reasoning. The company announced the new model, highlighting its ability to comprehend on-screen content, execute actions across various applications, and generate text with remarkable naturalness. This advanced capability comes with a significantly lower price point compared to its predecessors, making sophisticated AI-driven computer usage more accessible to a wider range of users and developers, as stated in its official announcement.

GPT-6.1 Terra Revolutionizes How Computers Are Used

Positioned as the successor to GPT-5.6 Terra, OpenAI’s new GPT-6.1 Terra is described as a fast and efficient model specifically engineered for computer use and visual reasoning. The company emphasizes its ability to understand what is displayed on a screen, perform operations across different applications, and produce naturally written language. This innovative AI is designed to streamline a wide array of daily tasks, from preparing business proposals and interpreting technical drawings to filling out forms and even editing videos, enhancing overall workday efficiency for its users.

Significant Cost Reductions Make AI More Accessible

One of the most compelling aspects of GPT-6.1 Terra is its dramatically reduced pricing structure. OpenAI reports that standard API costs for GPT-6.1 Terra are 50 percent lower than those for GPT-5.6 Terra and GPT-6.1 Sol. The model is priced at $1 per million input tokens, $0.05 per million cached input tokens, and $5 per million output tokens. This aggressive pricing strategy aims to make high-volume automation tasks, such as processing invoices and integrating data into accounting software, economically viable and practical for developers and businesses alike.

Performance Benchmarks Showed Remarkable Advancement

GPT-6.1 Terra has demonstrated superior performance in various internal tests. In the “Agents’ Last Exam,” which evaluates professional tasks, the model scored 15 percentage points higher than GPT-5.6 Terra at a medium reasoning level, all while incurring approximately one-third of the cost. Furthermore, OpenAI states that GPT-6.1 Terra achieves performance comparable to GPT-6 Astra at merely one-tenth of the price per token, completing tasks up to three times faster than both GPT-6 Astra and GPT-6.1 Sol. At its highest reasoning level, Terra reportedly offers a cost-per-task that is about nine times lower than that of its competitor, Anthropic’s Claude Opus 5.5 model.

Accuracy and Software Development Capabilities Have Significantly Improved

The new model also marks a substantial leap in accuracy and software development capabilities. OpenAI’s internal testing, which uses prompts designed to induce errors, revealed a drastic reduction in the error rate for GPT-6.1 Terra. At a low reasoning level, the error rate dropped from 53.6 percent in GPT-5.6 Terra to just 16.7 percent. For the highest reasoning level, the error rate further decreased to an impressive 4.3 percent. In the Terminal-Bench 4.0 test, which assesses the ability to resolve software issues from the command line, GPT-6.1 Terra solved 50.3 percent of tasks at the highest reasoning level, more than doubling GPT-5.6 Terra’s score of 21.5 percent. However, OpenAI acknowledges that for the most complex coding challenges, GPT-6.1 Sol and GPT-6 Astra remain the preferred options.

Broad Availability Ensures Wide Adoption Is Possible

GPT-6.1 Terra is now immediately accessible to a broad user base. ChatGPT Work and Codex users across Plus, Pro, Enterprise, and Edu tiers can utilize the new model starting today. Developers can integrate the model into their applications via the OpenAI API, using the identifier “gpt-6.1-terra.” According to OpenAI’s developer documentation, the model offers a substantial context window of 1,050,000 tokens and supports the computer use tool through its Responses API.

Robust Security Framework Protects Users Effectively

OpenAI has also placed a strong emphasis on the security aspects of GPT-6.1 Terra. The company confirms that the model meets the “Critical” threshold in cybersecurity under its Preparedness Framework, maintaining the same robust protections developed for GPT-6 Astra. Furthermore, extensive testing has demonstrated the model’s strong resilience against malicious instructions designed to divert users from their intended tasks, reinforcing its commitment to user safety and ethical AI deployment.

What do you think about AI models taking control of your computer? Share your thoughts and predictions for the future of human-computer interaction in the comments below!