Claude Sonnet 5

编程9 项

• ProgramBenchVals AI0.0%+
• Vibe Code Bench v1.1Vals AI81.3%+
• Code MigrationVals AI44.4%+
• LiveBench CodingLiveBench80.7%+
• CyberBench Patch v1.1Vals AI82.1%+
• Terminal-Bench 4.0Artificial Analysis14.1%+
• Vibe Code Bench 1–100Vals AI13.8%+
• SciCodeArtificial Analysis54.3%+
• Terminal-Bench 4.0Vals AI10.6%+

科研7 项

• Chess PuzzlesEpoch AI45.0%+
• LiveBench ReasoningLiveBench88.7%+
• Mystery Game PuzzlesEpoch AI35.0%+
• CritPtArtificial Analysis16.9%+
• FrontierMath T1–3Epoch AI65.6%+
• Humanity's Last ExamArtificial Analysis41.3%+
• Terminal-Bench ScienceVals AI5.7%+

专业办公9 项

• AA-Analyst AgentArtificial Analysis46.3%+
• APEX-AgentsMercor54.5%+
• AA-BriefcaseArtificial Analysis1,357.2+
• Finance AgentVals AI53.9%+
• GDPval-AAArtificial Analysis1,462.5+
• MedCodeVals AI47.5%+
• LiveBench Data AnalysisLiveBench71.7%+
• AutomationBench-AAArtificial Analysis36.5%+
• Harvey 法律 Agent 基准Vals AI5.0%+

知识问答7 项

• PBullshitBench v2BullshitBench78.0%+
• AA-OmniscienceArtificial Analysis16.5+
• AA-LCRArtificial Analysis82.0%+
• LiveBench LanguageLiveBench75.0%+
• LiveBench IFLiveBench63.9%+
• SimpleQA VerifiedEpoch AI33.7%+
• MLCR-AAArtificial Analysis5.0%+

真人盲选1 项

• Arena TextLMArena1,461.5+

尚无成绩的评测

这些评测尚无可采用的公开成绩，缺测不补分。