OpenAI just taught its model to use your computer better than most interns can.

The Summary

The Signal

GPT-6.1 Terra represents OpenAI's first model purpose-built for computer use, not just conversation. The model can navigate software interfaces, click buttons, fill forms, and execute multi-step workflows across applications. This is the agent economy arriving through the front door of every knowledge worker's desktop.

The timing matters. While Anthropic has been experimenting with computer use capabilities in Claude, OpenAI's entry with a dedicated model variant signals this is no longer a research curiosity. It's infrastructure. The "6.1" designation suggests this is an iteration on GPT-6, optimized for a specific use case rather than general reasoning.

"Fast and capable computer use" means agents that don't just understand what you want, they execute it across the actual software you already use.

The Hacker News reaction shows developer appetite, not skepticism. 288 points and 148 comments suggest people are already thinking through implementations. The question isn't whether AI agents will automate workflows. It's whose agents, on what terms, with what level of user control.

The Implication

If you're building software, you now have to assume an AI agent will be using it alongside humans. That changes interface design, API priorities, and what "user experience" even means. If you're a knowledge worker, the boring parts of your job just became automatable at scale, which means your value is about to compress into the parts that require judgment, taste, or relationships.

Watch how OpenAI prices this. If Terra is bundled into ChatGPT Plus, agents become consumer infrastructure overnight. If it's a separate tier, we're still in the early enterprise adoption phase. Either way, the shift from "AI that helps" to "AI that does" just accelerated.

Sources

Hacker News Best | OpenAI Blog