AI agent teams waste massive tokens for barely measurable quality gains, research finds
Multi-agent teams cost 5.1x more for barely measurable gains. One in four tests showed real improvement.
Why it matters
Enterprise teams deploying multi-agent orchestration face a hard ROI ceiling: scaling agent teams dramatically increases token spend without proportional quality gains. Anthropic's own data shows quality plateaus beyond ten agents—a critical finding for practitioners budgeting agentic workflows.
The key facts
5 to knowMulti-agent teams cost up to 5.1x more tokens than solo agents
Only 1 of 4 tests (25%) showed measurable quality improvement with GPT-6 Sol and Claude Opus 5.5
Anthropic internal data: quality plateaus beyond 10 agents while token costs continue climbing
Source: Vals AI research; corroborated by Anthropic's own eval data
Implication: token efficiency sharply degrades as team size scales
The story so far
Earlier coverage of this storyline
Go to the source
The Decoderthe-decoder.com
Publisher excerpt: Teams of AI agents barely outperform solo agents but cost up to 5.1x more, according to Vals AI. Only one out of four tests with GPT-6 Sol and Claude Opus 5.5 showed a measurable gain. Anthropic's own data backs this up: beyond ten agents, quality plateaus while token costs keep climbing.