Qwen3.8 Max 0902

编程61.611 项

• Terminal-Bench 4.0Vals AI38.9%折算分 67.1+
• Terminal-Bench 4.0Artificial Analysis38.9%折算分 66.7+
• Code Migration v1.1Vals AI57.2%折算分 64.7+
• MirrorCodeEpoch AI47.8%折算分 63.9+
• FrontierCode 2.0Artificial Analysis23.4%折算分 63.5+
• Vibe Code Bench v1.1Vals AI77.1%折算分 62.7+
• TapTap MakerTapTap Maker76.2%折算分 62.3+
• TTerminal-Bench 4.0Terminal-Bench27.0%折算分 60.2+
• PFrontierSWE v2Proximal Labs20.8%折算分 55.2+
• SciCodeArtificial Analysis52.1%折算分 52.9+
• Vibe Code Bench 1–100Vals AI24.8%+

科研61.36 项

• CritPtArtificial Analysis17.7%折算分 63.0+
• Terminal-Bench ScienceArtificial Analysis11.9%折算分 62.6+
• Humanity's Last ExamArtificial Analysis43.1%折算分 60.4+
• FrontierMath T1–3Epoch AI65.6%折算分 59.6+
• Chess PuzzlesEpoch AI37.0%折算分 59.3+
• CritPt v2 HardArtificial Analysis29.5%折算分 58.9+

专业办公67.59 项

• TaxAgentBenchVals AI49.7%折算分 79.1+
• AA-BriefcaseArtificial Analysis1,606.1折算分 72.0+
• Harvey 法律 Agent 基准 v1.1Vals AI15.4%折算分 71.1+
• GDPval-AAArtificial Analysis1,673.9折算分 69.3+
• Legal Research BenchVals AI46.2%折算分 68.5+
• Excel Modeling v1.1Vals AI72.0%折算分 67.9+
• AA-Analyst AgentArtificial Analysis45.0%折算分 66.9+
• AutomationBench-AAArtificial Analysis56.2%折算分 64.7+
• Finance Agent v2.1Vals AI58.8%折算分 62.0+

知识问答60.64 项

• MLCR-AAArtificial Analysis20.0%折算分 64.0+
• AA-Omniscience v1.1Artificial Analysis10.5折算分 57.5+
• SimpleQA VerifiedEpoch AI47.3%折算分 57.4+
• AA-LCRArtificial Analysis80.3%折算分 54.3+

真人盲选65.61 项

• Arena WebDevLMArena1,675.9折算分 69.7+

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。