Benchmark Registry

Alibaba (Tongyi)

Benchmark results for 8 Alibaba (Tongyi) models from official publications. Updated 2026-10-07.

Established
Not available
Latest model
Qwen3.8-Max (August 2026)

Benchmarks

161 results

ModelBenchmarkScoreSourceRegistry No.
Qwen3.5-397B-A17BMedXpertQA MM — Mercor single-shot65.8%mercor.comArchived copyAccuracyReported Benchmark leaderboardMercorIndependentRecord130003
Qwen3.5-397B-A17B (none)GPQA Diamond — Epoch AI harness86.4%epoch.aiArchived copyGPQA Diamond AccuracyReported Evaluated Benchmark leaderboardEpoch AIIndependentRecord130003
Qwen3.5-397B-A17BGPQA Diamond — Epoch AI harness85.9%epoch.aiArchived copyGPQA Diamond AccuracyReported Evaluated Benchmark leaderboardEpoch AIIndependentRecord130003
Qwen3.5-397B-A17BHumanity's Last Exam Text only — Mercor single-shot29.2%mercor.comArchived copyHumanity's Last Exam AccuracyReported Benchmark leaderboardMercorIndependentRecord130003
Qwen3.5-397B-A17BMMLU-Pro Original — Mercor single-shot88.0%mercor.comArchived copyAccuracyReported Benchmark leaderboardMercorIndependentRecord130003
Qwen3.5-397B-A17BSWE-bench Verified — Mercor mini-swe-agent, 1000 steps72.1%mercor.comArchived copyResolved TasksReported Benchmark leaderboardMercorIndependentRecord130003
Qwen3.5-397B-A17BSWE-bench Multilingual — Mercor mini-swe-agent, 1000 steps61.7%mercor.comArchived copyResolved TasksReported Benchmark leaderboardMercorIndependentRecord130003
Qwen3.5-397B-A17BAPEX-Agents Original24.9%mercor.comArchived copyPass rateReported Benchmark leaderboardMercorIndependentRecord130003
Qwen3.5-397B-A17BTerminal-Bench 2.1 — Mercor Terminus 253.9%mercor.comArchived copyTerminal-Bench AccuracyReported Benchmark leaderboardMercorIndependentRecord130003
Qwen3.5-397B-A17BCharXiv Descriptive and reasoning — Mercor single-shot91.6%mercor.comArchived copyCharXiv AccuracyReported Benchmark leaderboardMercorIndependentRecord130003
Qwen3.5-397B-A17BDeepSWE 1.1 — Mercor mini-swe-agent, 500 steps0.0%mercor.comArchived copyDeepSWE Pass@1Reported Benchmark leaderboardMercorIndependentRecord130003
Qwen3.6-PlusSWE-bench Verified — Epoch AI harness57.9%epoch.aiArchived copyResolved TasksReported Evaluated Benchmark leaderboardEpoch AIIndependentRecord130004
Qwen3.6-Plus (high)AutomationBench 1.0.64.3%zapier.comArchived copyTask pass rateReported Benchmark leaderboardZapierIndependentZapier AutomationBench TeamRecord130004
Qwen3.6-PlusGPQA Diamond — Epoch AI harness88.4%epoch.aiArchived copyGPQA Diamond AccuracyReported Evaluated Benchmark leaderboardEpoch AIIndependentRecord130004
Qwen3.6-PlusSimpleQA Verified Original — Epoch AI harness44.1%epoch.aiArchived copyAccuracyReported Evaluated Benchmark leaderboardEpoch AIIndependentRecord130004
Qwen3.6-35B-A3B (none)GPQA Diamond — Epoch AI harness84.8%epoch.aiArchived copyGPQA Diamond AccuracyReported Evaluated Benchmark leaderboardEpoch AIIndependentRecord130005
Qwen3.6-35B-A3BGPQA Diamond — Epoch AI harness83.8%epoch.aiArchived copyGPQA Diamond AccuracyReported Evaluated Benchmark leaderboardEpoch AIIndependentRecord130005
Qwen3.7-MaxGPQA Diamond — Epoch AI harness90.9%epoch.aiArchived copyGPQA Diamond AccuracyReported Evaluated Benchmark leaderboardEpoch AIIndependentRecord130006
Qwen3.7-Max (Thinking on)Riemann-bench October 2026 leaderboard15.2%surgehq.aiArchived copyRiemann-bench Pass@1Reported Benchmark leaderboardSurge AIIndependentRecord130006
Qwen3.7-MaxSWE-bench Verified — Epoch AI harness77.3%epoch.aiArchived copyResolved TasksReported Evaluated Benchmark leaderboardEpoch AIIndependentRecord130006
Qwen3.7-MaxSimpleQA Verified Original — Epoch AI harness55.8%epoch.aiArchived copyAccuracyReported Evaluated Benchmark leaderboardEpoch AIIndependentRecord130006
Qwen3.7-Plus (none)GPQA Diamond — Epoch AI harness81.8%epoch.aiArchived copyGPQA Diamond AccuracyReported Evaluated Benchmark leaderboardEpoch AIIndependentRecord130007
Qwen3.7-PlusAutomationBench 1.0.63.7%zapier.comArchived copyTask pass rateReported Benchmark leaderboardZapierIndependentZapier AutomationBench TeamRecord130007
Qwen3.7-PlusGPQA Diamond — Epoch AI harness87.9%epoch.aiArchived copyGPQA Diamond AccuracyReported Evaluated Benchmark leaderboardEpoch AIIndependentRecord130007
Qwen3.7-Plus (Thinking on)Chartography October 2026 leaderboard15.8%surgehq.aiArchived copyChartography Pass@1Reported Benchmark leaderboardSurge AIIndependentRecord130007
Qwen3.8-Max (xhigh)CharXiv Descriptive and reasoning — Mercor single-shot94.1%mercor.comArchived copyCharXiv AccuracyReported Benchmark leaderboardMercorIndependentRecord130008
Qwen3.8-Max (xhigh)DeepSWE 1.1 — Mercor mini-swe-agent, 500 steps53.1%mercor.comArchived copyDeepSWE Pass@1Reported Benchmark leaderboardMercorIndependentRecord130008
Qwen3.8-Max (xhigh)BrowseComp 130-question subset — Mercor web research agent67.9%mercor.comArchived copyBrowseComp AccuracyReported Benchmark leaderboardMercorIndependentRecord130008
Qwen3.8-Max (xHigh)Riemann-bench October 2026 leaderboard15.2%surgehq.aiArchived copyRiemann-bench Pass@1Reported Benchmark leaderboardSurge AIIndependentRecord130008
Qwen3.8-Max (xhigh)SimpleQA Verified Original — Epoch AI harness45.8%epoch.aiArchived copyAccuracyReported Evaluated Benchmark leaderboardEpoch AIIndependentRecord130008
Qwen3.8-Max (xhigh)Humanity's Last Exam Text only — Mercor single-shot42.2%mercor.comArchived copyHumanity's Last Exam AccuracyReported Benchmark leaderboardMercorIndependentRecord130008
Qwen3.8-MaxFrontierSWE 2 — October 2026 evaluation15.8%frontierswe.comArchived copyFrontierSWE ScoreReported Benchmark leaderboardProximalIndependentRecord130008
Qwen3.8-Max (xhigh)MMLU-Pro Original — Mercor single-shot88.8%mercor.comArchived copyAccuracyReported Benchmark leaderboardMercorIndependentRecord130008
Qwen3.8-Max (xhigh)MedXpertQA MM — Mercor single-shot78.3%mercor.comArchived copyAccuracyReported Benchmark leaderboardMercorIndependentRecord130008
Qwen3.8-Max (xHigh)Chartography October 2026 leaderboard29.1%surgehq.aiArchived copyChartography Pass@1Reported Benchmark leaderboardSurge AIIndependentRecord130008
Qwen3.8-Max (xhigh)SWE-bench Verified — Mercor mini-swe-agent, 1000 steps83.6%mercor.comArchived copyResolved TasksReported Benchmark leaderboardMercorIndependentRecord130008
Qwen3.8-Max (xhigh)SWE Atlas Codebase QnA — Mercor mini-swe-agent, 250 steps50.5%mercor.comArchived copyMean pass@1Reported Benchmark leaderboardMercorIndependentRecord130008
Qwen3.8-Max (xhigh)GPQA Diamond — Mercor single-shot92.5%mercor.comArchived copyGPQA Diamond AccuracyReported Benchmark leaderboardMercorIndependentRecord130008
Qwen3.8-Max (xhigh)SWE-bench Multilingual — Mercor mini-swe-agent, 1000 steps88.3%mercor.comArchived copyResolved TasksReported Benchmark leaderboardMercorIndependentRecord130008
Qwen3.8-Max (xhigh)GPQA Diamond — Epoch AI harness92.7%epoch.aiArchived copyGPQA Diamond AccuracyReported Evaluated Benchmark leaderboardEpoch AIIndependentRecord130008
Qwen3.8-Max (xhigh)MMMU-Pro 1,730 questions — Mercor single-shot82.8%mercor.comArchived copyMMMU-Pro accuracySetting: Not statedReported Benchmark leaderboardMercorIndependentRecord130008
Qwen3.8-Max (xhigh)APEX-Agents Original63.3%mercor.comArchived copyPass rateReported Benchmark leaderboardMercorIndependentRecord130008
Qwen3.6-PlusHLAB October 2026 evaluation1.3%vals.aiArchived copyHLAB Task Pass RateReported Benchmark leaderboardVals AIIndependentRecord130004
Qwen3.6-PlusFinance Agent Benchmark 2 — October 2026 evaluation40.8%vals.aiArchived copyFinance Agent AccuracyReported Benchmark leaderboardVals AIIndependentRecord130004
Qwen3.6-PlusVibe Code Bench 1.1 — OpenHands, October 2026 evaluation25.6%vals.aiArchived copyVibe Code Bench ScoreReported Benchmark leaderboardVals AIIndependentRecord130004
Qwen3.7-MaxHLAB October 2026 evaluation1.7%vals.aiArchived copyHLAB Task Pass RateReported Benchmark leaderboardVals AIIndependentRecord130006
Qwen3.7-MaxVibe Code Bench 1.1 — OpenHands, October 2026 evaluation47.7%vals.aiArchived copyVibe Code Bench ScoreReported Benchmark leaderboardVals AIIndependentRecord130006
Qwen3.7-MaxFinance Agent Benchmark 2 — October 2026 evaluation47.8%vals.aiArchived copyFinance Agent AccuracyReported Benchmark leaderboardVals AIIndependentRecord130006
Qwen3.7-PlusHLAB October 2026 evaluation0.0%vals.aiArchived copyHLAB Task Pass RateReported Benchmark leaderboardVals AIIndependentRecord130007
Qwen3.7-PlusVibe Code Bench 1.1 — OpenHands, October 2026 evaluation46.4%vals.aiArchived copyVibe Code Bench ScoreReported Benchmark leaderboardVals AIIndependentRecord130007

Alibaba (Tongyi) models