Anthropic's Claude Managed Agents can now split a task across up to 1,000 sub-agents running in parallel

Anthropic has added dynamic workflows to Claude Managed Agents, its hosted agent infrastructure. A lead agent writes a plan, hands tasks to sub-agents, and merges their results when they finish. Each execution can run up to 1,000 agents in parallel. Anthropic tested the feature by hiding 70 bugs in a 116,000-line codebase. A single agent found between 14 and 27 bugs per run, while the dynamic workflow consistently found 66, according to The Decoder. Whether those gains carry over to other kinds of tasks has not been shown. Cost is a concern: a senior OpenAI engineer recently called agent swarms a massive waste of tokens, and Anthropic itself warns that the workflows can use 'a lot of tokens' and recommends starting small. Developers turn on the feature by choosing the 'multiagent_20261001' agent type. They can follow the documentation or run '/claude-api managed-agents-onboard' in Claude Code to get started. The launch gives enterprise developers a managed option for large multi-agent jobs, but teams will need to measure the cost and quality trade-offs on their own workloads.