New Delhi: OpenAI has launched GPT-6.1 Terra, a new artificial intelligence model designed to handle computer-based tasks, understand visual information and generate natural, clear writing. The model is intended to go beyond conventional text-based interactions by allowing AI agents to work across applications and complete multi-step tasks.
According to OpenAI, GPT-6.1 Terra can interpret what appears on a computer screen and take actions within software, making it suitable for tasks such as filling out online forms, processing invoices, preparing business proposals and working with technical drawings. The company announced the model on October 26, highlighting its computer-use capabilities and lower operating costs.
GPT-6.1 Terra is available to Plus, Pro, Enterprise and Edu users through ChatGPT Work and Codex. Developers can also access it through the OpenAI API under the model name gpt-6.1-terra. The company says the model’s standard API pricing is 50% lower than that of GPT-5.6 Terra and GPT-6.1 Sol.
What Can GPT-6.1 Terra Do?
Unlike models focused primarily on generating text, GPT-6.1 Terra is designed to interact with software interfaces. It can interpret information displayed on a screen and use computer controls to carry out tasks across applications.
Potential uses include reading invoice details and entering them into accounting software, completing online forms, drafting proposals and analysing technical drawings. OpenAI also highlighted tasks involving the review of animated film footage and the creation of time-coded logs for restoration work.
The model’s performance was evaluated using benchmarks that test how well AI systems carry out professional tasks and interact with desktop applications. OpenAI said GPT-6.1 Terra scored 15 percentage points higher than GPT-5.6 Terra on the Agents’ Last Exam benchmark at medium reasoning effort, while operating at about one-third of the cost.
OpenAI Reports Lower Error Rates
OpenAI has also highlighted improvements in factual accuracy. In its internal evaluation using prompts designed to trigger errors, GPT-6.1 Terra recorded an error rate of 16.7% at low reasoning effort, compared with 53.6% for GPT-5.6 Terra.
At maximum reasoning effort, the reported error rate fell further to 4.3%, an improvement of 18 percentage points over the earlier model at the corresponding setting. These figures come from OpenAI’s own testing and should not be interpreted as a guarantee that the model will make no mistakes in everyday use.
The company has also reported gains in coding performance. On Terminal-Bench 4.0, which tests how well AI agents solve command-line tasks, GPT-6.1 Terra achieved a score of 50.3% at maximum reasoning effort. That was more than double the published 21.5% score for GPT-5.6 Terra.
OpenAI said GPT-6.1 Terra is designed for routine professional work and high-volume tasks where speed and cost matter. For the most demanding reasoning and coding work, the company continues to position GPT-6.1 Sol and GPT-6 Astra as options for more complex assignments.
How Much Does GPT-6.1 Terra Cost?
OpenAI has set standard API prices at $1 per million input tokens and $5 per million output tokens. Cached input tokens cost $0.05 per million. The company says the lower pricing is intended to make it more economical for developers to automate large volumes of computer-based work.
With its combination of visual reasoning, computer interaction, coding improvements and lower API costs, GPT-6.1 Terra is positioned as a model for automating everyday professional workflows. However, as with other AI systems, its outputs and actions still need appropriate oversight, particularly when tasks involve sensitive information or consequential decisions.
