Muse Spark 1.3

编程9 项

• ProgramBenchVals AI2.5%+
• Vibe Code Bench v1.1Vals AI85.9%+
• Terminal-Bench 4.0Artificial Analysis33.3%+
• Code MigrationVals AI47.4%+
• Vibe Code Bench 1–100Vals AI20.5%+
• Terminal-Bench 4.0Vals AI24.7%+
• LiveBench CodingLiveBench81.1%+
• SciCodeArtificial Analysis58.8%+
• CyberBench Patch v1.1Vals AI82.1%+

科研9 项

• MysteryMechanismVals AI36.0%+
• CritPtArtificial Analysis24.9%+
• LiveBench ReasoningLiveBench89.7%+
• Humanity's Last ExamArtificial Analysis48.7%+
• FrontierMath T1–3Epoch AI74.0%+
• Terminal-Bench ScienceVals AI10.0%+
• Terminal-Bench ScienceArtificial Analysis11.0%+
• Chess PuzzlesEpoch AI37.0%+
• Mystery Game PuzzlesEpoch AI25.0%+

专业办公9 项

• Harvey 法律 Agent 基准Vals AI23.8%+
• τ³-BankingMercor52.2%+
• AA-BriefcaseArtificial Analysis1,581.3+
• GDPval-AAArtificial Analysis1,681+
• Finance AgentVals AI60.0%+
• LiveBench Data AnalysisLiveBench79.6%+
• AutomationBench-AAArtificial Analysis57.9%+
• AA-Analyst AgentArtificial Analysis37.5%+
• APEX-AgentsMercor47.6%+

知识问答5 项

• LiveBench IFLiveBench78.0%+
• AA-OmniscienceArtificial Analysis25+
• LiveBench LanguageLiveBench82.8%+
• AA-LCRArtificial Analysis83.0%+
• MLCR-AAArtificial Analysis13.9%+

真人盲选1 项

• Arena TextLMArena1,495.6+

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。