OpenAI has unveiled its latest artificial intelligence model, GPT-6.1 Terra, which launched on October 26, 2026. Rather than functioning solely as a chatbot that responds to questions, the new model is designed to act as an active digital assistant capable of viewing a computer screen, clicking on elements and carrying out everyday tasks.
Marking the launch on X, formerly Twitter, OpenAI CEO Sam Altman wrote, "6.1 Terra is here! It is good at a lot of things, but we are especially excited for this new era of computer use."
How GPT-6.1 Terra works on your computer
Unlike earlier AI models that primarily generate text, GPT-6.1 Terra is designed to interact directly with computer applications. It can interpret what is displayed on the screen and navigate software interfaces much like a human user. The model can click, type and move between applications to complete tasks involving multiple steps.
Businesses, for instance, could use it to transfer invoice information into accounting software, complete lengthy online forms, create business proposals or convert architectural drawings into 3D models. The aim is to allow AI to take care of repetitive and time-consuming digital work with minimal human intervention.
Faster performance and lower costs
GPT-6.1 Terra represents a significant improvement over the previous GPT-5.6 Terra. OpenAI says the new model completes tasks roughly three times faster on average than larger, more advanced models such as GPT-6.1 Sol and GPT-6 Astra. It is also more affordable for developers building applications around the technology.
The company has reduced standard API pricing by 50% compared with the previous version. The service now costs $1 per million input tokens and $5 per million output tokens, making it a more economical option for businesses that operate automated workflows at scale.
Big improvements in accuracy and coding
OpenAI has also focused on improving Terra's accuracy and reducing the number of mistakes it makes. According to the company's internal evaluations, the error rate on simple tasks dropped from 53.6% with the older model to 16.7% with GPT-6.1 Terra. When operating at its highest reasoning setting, the error rate declined further to 4.3%.
The model has also made substantial gains in coding, achieving more than twice the score of its predecessor on complex terminal benchmarks. Following these improvements, GPT-6.1 Terra is being introduced to developers through platforms such as GitHub Copilot.
Where does Terra fit in OpenAI's lineup?
OpenAI offers a range of AI models, including Luna, Terra, Sol and Astra. Luna is intended for lightweight tasks, while Sol and Astra cater to more demanding and complex workloads. Terra occupies the middle ground, offering a balance between capability, speed and cost.
After last month's updates to Sol and Luna, the latest release brings Terra into the GPT-6.1 generation.
GPT-6.1 Terra is now available to individual users and businesses through ChatGPT Plus, Pro, Enterprise and Edu subscriptions, as well as through ChatGPT Work and Codex. Developers can also access the model through OpenAI's API and Amazon Bedrock.
