Anthropic says single-agent setups often beat multiple sub-agents
What happened
On 25 October 2026, Anthropic published the view that a single agent loading context on demand often outperforms multi-sub-agent architectures on quality, cost and latency, and released a migration toolkit to help teams assess whether to consolidate their sub-agents. Anthropic says models from Opus 4.5 onwards can manage complex plans directly in a single loop; in one comparison, the single-agent setup was about twice as fast as the multi-agent approach. So far the development rests on Anthropic's conclusions and the accompanying migration tool, with no reports of third-party reproduction or independent evaluation.
Written by AI from the reports · updated 2 h ago
Timeline
Follow the reports to see the event from each side.
- The DecoderAnthropic says one AI agent beats many subagents now that models can handle complexity alone
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.
Heat of this event
Not enough continuous observations yet to draw a trend.