Story

arxiv_agent_systems_research ยท Nov 13, 2026 ยท paper

Source brief

More Agents, More Samples, or More Thinking? Untangling Coordination, Ensembling, and Test-Time Compute in Multi-Agent LLMs

arxiv.orgNov 13, 2026
original source linked

In brief

Evaluations of multi-agent language models often compare communicating architectures against a single greedy pass. This practice confounds coordination with sample multiplicity and inference compute. We synthesize nin...

Feed lens
agentevaluation

Continue reading

Read the original at arxiv.org โ†’Open in live feedRead that dayโ€™s brief

Earlier in this thread 4 items