Anthropic says one agent often outperforms multi-agent setups
Original title: Anthropic says one AI agent beats many subagents now that models can handle complexity alone
The firm is publishing a GitHub migration kit and says one comparison ran about twice as fast.
Anthropic's Applied AI team says a single primary agent that loads context on demand often beat specialized sub-agent fleets on quality, cost, and latency across enterprise deployments. Newer models from Opus 4.5 onward can manage complex plans in one loop. The company published a GitHub migration kit and warns against merging blindly after one support-agent test reduced latency but broke a working process.
Unconfirmed: Speed, quality, and cost comparisons are Anthropic's own deployment findings, not independent benchmarks.
- Claude Code's system prompt for newer models was cut by over 80 percent.
- Tool selection tends to degrade beyond 30 to 50 tool definitions.
- Sub-agents still fit parallel tasks or when context-window performance drops.
- A Vals study found larger agent teams raised cost without quality gains.
- One customer had a dozen sub-agents inside a customer-facing agent.
- Vendors
- Anthropic
- Products
- Claude Code
From the original article
3 min read at the-decoder.com
Anthropic says companies should evaluate whether they can consolidate sub-agents into a single agent that loads context on demand. In one architecture comparison, the single-agent setup ran about twice as fast. Anthropic has published a migration kit on GitHub to help teams make the transition.
By Matthias Bastian · the-decoder.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.