ResearchBenchmarksMulti-hop Fact Verification on HoVer 3-HopFollow58Macro F1MERMAID48.6451.0753.555.93Jan 29, 2026Evaluation ResultsMethodMethodLinksMacro F1MERMAIDLLM=GPT-5 MiniLLM=GPT-5 Mini2026.0158MERMAIDLLM=GPT-4oLLM=GPT-4o2026.0156FOLK2026.0155MERMAIDLLM=Qwen-2.5-70BLLM=Qwen-2.5-70B2026.0152ProgramFC2026.0151MERMAIDLLM=OSS-120BLLM=OSS-120B2026.0150Self-Ask2026.0149