SwarmBench: Can Large Language Models Act as Agent Swarm Orchestrators?
Large language model-based multi-agent systems are evolving from fixed interaction topologies toward dynamically orchestrated Agent Swarms. However, existing benchmarks are still largely based on single-agent or general-purpose agent tasks, making it difficult to systematically evaluate key orchestration capabilities. We propose SwarmBench, a benchmark that evaluates model performance from multiple perspectives, including accuracy, efficiency, cost, and process quality. Experimental results show that current models exhibit substantial differences in orchestration capability. These differences
Lineage graph
Paper → model → repo connections mined from source citations (Tier-1 exact match).
Why these links exist
Every edge carries a method, confidence, and the source snippet that justified it — so bad links are debuggable.
- FuzzySimilar title/name (fuzzy) · 59%AgentCore-8B →
“Fuzzy title match (0.73): “SwarmBench: Can Large Language Models Act as Agent Swarm Orc” ≈ “AgentCore-8B””
- PossiblePossibly related (embedding) · 28%deepset-ai/haystack →
“Possibly related via embedding similarity 0.63 (not asserted). Timestamp check: artifact slightly before paper (-60d).”
- PossiblePossibly related (embedding) · 61%Capable language models can outgrow the benefits of collaboration →
- PossiblePossibly related (embedding) · 50%Project HydraFusion: Frontier quality via multi-model orchestration →
- FuzzySimilar title/name (fuzzy) · 87%SWE-agent/SWE-agent →
“Fuzzy title match (0.94): “SwarmBench: Can Large Language Models Act as Agent Swarm Orc” ≈ “SWE-agent/SWE-agent””
- FuzzySimilar title/name (fuzzy) · 87%zhayujie/CowAgent →
“Fuzzy title match (0.94): “SwarmBench: Can Large Language Models Act as Agent Swarm Orc” ≈ “zhayujie/CowAgent””
- FuzzyOverlapping authors or contributors · 62%keras-team/keras →
“Shared author/contributor keys: jin”
- FuzzyOverlapping authors or contributors · 62%HKUDS/LightRAG →
“Shared author/contributor keys: jin”