OpenAI Launches GPT-6.1 Terra With 50% Lower API Costs And AI That Controls Apps

OpenAI's new GPT-6.1 Terra isn't interested in just answering questions. It wants to get the work done, too!

Launched on October 26, 2026, the new artificial intelligence (AI) model can interact with computer screens, navigate applications, and execute multistep tasks, while offering lower operating costs and improved accuracy over its predecessor.
 

TL;DR

 

• OpenAI launched GPT-6.1 Terra with computer-use capabilities, allowing AI to click, type, and navigate applications.
• The model reportedly delivers three times faster task execution than larger models in OpenAI's lineup.
• API pricing has dropped by 50% compared to GPT-5.6 Terra.
• GPT-6.1 Terra is available across supported ChatGPT subscriptions, Codex, OpenAI's API, and Amazon Bedrock.

 

OpenAI's GPT-6.1 Terra Can Operate Apps And Automate Workflows

OpenAI is pushing Terra beyond conventional chatbot interactions by allowing the model to visually interpret computer screens and perform actions through software interfaces.

The model can move between applications, enter information, click interface elements, and execute sequences of actions instead of simply explaining how users should complete them.

According to reports, potential applications include transferring invoice information into accounting systems, completing online forms, preparing proposals, and converting architectural drawings into 3D models.

The capabilities extend OpenAI's push toward autonomous digital workflows, following its launch of ChatGPT Work, which brought AI-powered project execution across files and applications.
 

GPT-6.1 Terra Cuts API Pricing By 50% And Improves Accuracy

Alongside enhanced computer-use capabilities, OpenAI is targeting two persistent enterprise AI challenges: operating costs and execution errors.

GPT-6.1 Terra reportedly completes tasks around three times faster on average than heavier models such as GPT-6.1 Sol and GPT-6 Astra.

Its standard API pricing is also 50% lower than its predecessor, dropping to $1 per million input tokens and $5 per million output tokens.

The reduction follows OpenAI's earlier GPT-5.6 pricing adjustments, which lowered Terra's API costs by 20% and Luna's by 80%.

However, improved affordability isn't the only significant development.

OpenAI's internal evaluations reportedly show that GPT-6.1 Terra reduced its error rate on simple tasks from 53.6% with the previous generation to 16.7%. At maximum reasoning effort, that figure reportedly declined further to 4.3%.

The model also achieved more than twice its predecessor's performance on complex terminal coding benchmarks, with availability expanding to developers through GitHub Copilot.

Together, these improvements position Terra as an option for organizations seeking greater automation without relying exclusively on OpenAI's most computationally demanding models.
 

 

GPT-6.1 Terra Availability Expands Across ChatGPT, Codex And Amazon Bedrock

GPT-6.1 Terra occupies the middle tier of OpenAI's model portfolio, positioned between the lightweight Luna model and the more advanced Sol and Astra offerings.

The model is reportedly available through ChatGPT Plus, Pro, Enterprise, and Edu subscriptions within ChatGPT Work and Codex, alongside the OpenAI API and Amazon Bedrock.

Its launch follows the earlier GPT-5.6 model family rollout, which established separate options for speed, affordability, and advanced reasoning.

With GPT-6.1 Terra, OpenAI is now bringing faster execution, lower costs, and improved computer interaction to its balanced AI offering.

The development could make application-driven automation more practical for businesses, although real-world reliability will ultimately determine how extensively organizations adopt its computer-use capabilities.