List your company

Research

Benchmarks

Compare state-of-the-art methods across the tasks and datasets used to measure AI progress.

309,131 benchmarksPage 964 of 12,366

Benchmarks

TaskDatasetDomainTrendResultsLast update
Information RetrievalWIKI-BALANCEretrieval search rag28Feb 26, 2026
Information RetrievalWIKI-BALANCE (test)retrieval search rag28Feb 26, 2026
Malicious AgentCSQAadversarial safety securityagents tool use28Feb 27, 2026
Knowledge Poisoning AttackHotpotQAadversarial safety securitynlp28Jun 11, 2026
Jailbreak Attack SuccessSpatialJB Summary 1.0adversarial safety securityllm chat instruction28Feb 26, 2026
Code vulnerability repairLiveCVEBenchcode developer aiadversarial safety security28Feb 26, 2026
Backdoor DetectionBalanced 50% clean, 50% backdoored (test)adversarial safety security28Feb 26, 2026
Role-playingRoleLLMllm chat instruction28Feb 26, 2026
Empathetic DialogueKardiaBenchllm chat instructionnlp28Feb 26, 2026
Multiple Choice Question AnsweringArc-Cnlp28May 15, 2026
Natural Language UnderstandingGLUEnlp28Feb 26, 2026
Mathematical ReasoningMATH-500llm chat instructionnlp28Mar 24, 2026
Mathematical ReasoningOMEGAllm chat instructionnlp28Feb 27, 2026
Zero-shot Learning7 Downstream Tasks Avgcomputer visionnlp28Mar 27, 2026
ReasoningLiveCodeBench v6llm chat instructionnlp28Jul 7, 2026
Math ReasoningAIME25llm chat instructionnlp28Apr 24, 2026
Hallucination DetectionSQuADllm chat instructionnlp28May 12, 2026
Role-playingCharacterBench 1.0 (test)llm chat instruction28Feb 26, 2026
Memory-Augmented DialoguePersonaMem v1.0 (test)llm chat instructionnlp28Feb 26, 2026
Belief assessment on misinformationMISBELIEF Sports domain (Hard split)nlp28Feb 26, 2026
Belief assessment on misinformationMISBELIEF Sports domain Mediumnlp28Feb 26, 2026
Belief assessment on misinformationMISBELIEF Sports domain Easy splitnlp28Feb 26, 2026
Belief assessment on misinformationMISBELIEF Health domain Hardnlp28Feb 26, 2026
Belief assessment on misinformationMISBELIEF Health domain (Medium)nlp28Feb 26, 2026
Belief assessment on misinformationMISBELIEF Health domain Easy splitnlp28Feb 26, 2026