← Tuesday, July 28, 2026
AI Frontier · Jul 24
ModerateAnthropic's new Claude Opus 5 can run AI agents for hours
Anthropic put out a new AI model called Claude Opus 5. It is built to run agents, which are AIs that do a task on their own. This one can keep working for hours, even overnight. When it hits a snag, it can fix its own mistake and keep going. It costs the same as the last Opus model but is smarter.
Why it matters to GCs
More building firms now hand small jobs to AI agents, like sorting bids or checking submittals. A model that works longer and fixes its own errors needs less babysitting. That means an agent can finish a bigger job before a person has to step in.
VentureBeat·Via POLR AI Hub