Grok 4.6

编程59.913 项

• ProgramBenchVals AI1.0%折算分 至多 80.3+
• APEX-SWEMercor57.7%折算分 68.8+
• Vibe Code Bench v1.1Vals AI76.2%折算分 62.2+
• LiveBench Agentic CodingLiveBench57.0%折算分 61.0+
• MirrorCodeEpoch AI41.1%折算分 60.8+
• PFrontierSWE v2Proximal Labs27.5%折算分 59.7+
• Terminal-Bench 4.0Artificial Analysis17.2%折算分 58.9+
• Terminal-Bench 4.0Vals AI17.2%折算分 57.1+
• LiveBench CodingLiveBench76.8%折算分 56.7+
• CyberBench Patch v1.1Vals AI80.4%折算分 55.9+
• TTerminal-Bench 4.0Terminal-Bench20.3%折算分 55.6+
• SciCodeArtificial Analysis53.0%折算分 54.1+
• Vibe Code Bench 1–100Vals AI14.8%+

科研62.911 项

• LiveBench ReasoningLiveBench90.5%折算分 66.7+
• MysteryMechanismVals AI30.6%折算分 66.7+
• BioMysteryBenchVals AI72.2%折算分 66.7+
• CritPtArtificial Analysis19.7%折算分 64.5+
• Terminal-Bench ScienceVals AI11.4%折算分 63.5+
• Mystery Game PuzzlesEpoch AI34.0%折算分 62.1+
• AARC-AGI-2ARC Prize67.1%折算分 61.8+
• Humanity's Last ExamArtificial Analysis44.1%折算分 61.2+
• FrontierMath T1–3Epoch AI66.0%折算分 59.8+
• Terminal-Bench ScienceArtificial Analysis6.2%折算分 57.0+
• Chess PuzzlesEpoch AI33.0%折算分 56.9+

专业办公65.810 项

• APEX-AgentsMercor65.3%折算分 71.4+
• AutomationBench-AAArtificial Analysis67.0%折算分 69.9+
• Legal Research BenchVals AI48.1%折算分 69.9+
• AA-BriefcaseArtificial Analysis1,541.9折算分 69.5+
• GDPval-AAArtificial Analysis1,654.2折算分 68.5+
• τ³-BankingMercor45.7%折算分 68.3+
• TaxAgentBenchVals AI33.1%折算分 67.3+
• AA-Analyst AgentArtificial Analysis41.3%折算分 64.1+
• LiveBench Data AnalysisLiveBench73.9%折算分 59.6+
• MedCodeVals AI44.7%折算分 55.5+

知识问答61.67 项

• LiveBench LanguageLiveBench83.7%折算分 65.8+
• PBullshitBench v2BullshitBench56.0%折算分 64.5+
• LiveBench IFLiveBench71.9%折算分 63.5+
• LNYT Connections 扩展版Lech Mazur80.0%折算分 60.0+
• MLCR-AAArtificial Analysis17.2%折算分 59.9+
• SimpleQA VerifiedEpoch AI48.9%折算分 58.8+
• AA-LCRArtificial Analysis81.0%折算分 55.2+

真人盲选59.92 项

• Arena WebDevLMArena1,616.4折算分 63.9+
• Arena TextLMArena1,455.5折算分 52.4+

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。