AI
Anthropic launches Claude Opus 5, calls it 'thoughtful and proactive' for long-running agents
Anthropic unveiled Claude Opus 5 on Tuesday, describing the new flagship as a step-change for long-running agents with state-of-the-art results on Frontier-Bench and GDPval-AA at roughly half the price of Claude Fable 5. Alongside the model the company introduced Claude Tag, an evolution of Claude Code in which tagging @Claude triggers proactive work; Anthropic said 65% of its product team's code now flows through the tool. A separate post documents Claude agents independently running an open-ended AI safety research project end-to-end.