Anthropic releases Opus 4.6 with new agent teams

Anthropic has released Claude 4.6 Opus, introducing a breakthrough feature called "Agent Teams." This update allows users to orchestrate multiple specialized agents that can communicate with each other to solve complex, multi-step problems.

Collaborative AI

Unlike traditional LLM interactions where one model handles everything, Agent Teams allows for a "manager" agent to delegate tasks to "specialist" agents (e.g., a coder, a researcher, and a reviewer). This collaborative approach significantly reduces errors and allows for much larger project scopes.

Anthropic's focus on safety and reliability remains at the core of Opus 4.6, with new "constitutional" guardrails that apply to the entire team's interactions. This release directly challenges OpenAI's recent agentic updates, setting the stage for a new era of multi-agent systems.