Benchmark Registry

Z.ai

Benchmark results for 9 Z.ai models from official publications. Updated 2026-10-07.

Established
Not available
Latest model
GLM-5.3-Flash (August 2026)

Benchmarks

174 results

ModelBenchmarkScoreSourceRegistry No.
GLM-4.7GPQA Diamond — Epoch AI harness83.3%epoch.aiArchived copyGPQA Diamond AccuracyReported Evaluated Benchmark leaderboardEpoch AIIndependentRecord150001
GLM-4.7SimpleQA Verified Original — Epoch AI harness32.2%epoch.aiArchived copyAccuracyReported Evaluated Benchmark leaderboardEpoch AIIndependentRecord150001
GLM-5DeepSWE 1.1 — Mercor mini-swe-agent, 500 steps2.1%mercor.comArchived copyDeepSWE Pass@1Reported Benchmark leaderboardMercorIndependentRecord150002
GLM-5GPQA Diamond — Epoch AI harness87.8%epoch.aiArchived copyGPQA Diamond AccuracyReported Evaluated Benchmark leaderboardEpoch AIIndependentRecord150002
GLM-5SWE-bench Verified — Epoch AI harness72.1%epoch.aiArchived copyResolved TasksReported Evaluated Benchmark leaderboardEpoch AIIndependentRecord150002
GLM-5SWE Atlas Refactoring — mini-swe-agent24.2%labs.scale.comArchived copyMean pass@1Reported Benchmark leaderboardScale AIIndependentRecord150002
GLM-5SWE Atlas Codebase QnA — Public; mini-swe-agent20.5%labs.scale.comArchived copyMean pass@1Reported Benchmark leaderboardScale AIIndependentRecord150002
GLM-5.1DeepSWE 1.1 — Mercor mini-swe-agent, 500 steps18.9%mercor.comArchived copyDeepSWE Pass@1Reported Benchmark leaderboardMercorIndependentRecord150003
GLM-5.1APEX-Agents Original40.9%mercor.comArchived copyPass rateReported Benchmark leaderboardMercorIndependentRecord150003
GLM-5.1SimpleQA Verified Original — Epoch AI harness34.0%epoch.aiArchived copyAccuracyReported Evaluated Benchmark leaderboardEpoch AIIndependentRecord150003
GLM-5.1SWE-bench Verified — Epoch AI harness74.2%epoch.aiArchived copyResolved TasksReported Evaluated Benchmark leaderboardEpoch AIIndependentRecord150003
GLM-5.1GPQA Diamond — Epoch AI harness89.9%epoch.aiArchived copyGPQA Diamond AccuracyReported Evaluated Benchmark leaderboardEpoch AIIndependentRecord150003
GLM-5.1Humanity's Last Exam Text only — Mercor single-shot29.4%mercor.comArchived copyHumanity's Last Exam AccuracyReported Benchmark leaderboardMercorIndependentRecord150003
GLM-5.1 (max)AutomationBench 1.0.66.2%zapier.comArchived copyTask pass rateReported Benchmark leaderboardZapierIndependentZapier AutomationBench TeamRecord150003
GLM-5.2 (max)SWE-bench Multilingual — Mercor mini-swe-agent, 1000 steps78.6%mercor.comArchived copyResolved TasksReported Benchmark leaderboardMercorIndependentRecord150004
GLM-5.2 (Max)Riemann-bench October 2026 leaderboard10.4%surgehq.aiArchived copyRiemann-bench Pass@1Reported Benchmark leaderboardSurge AIIndependentRecord150004
GLM-5.2 (max)SimpleQA Verified Original — Epoch AI harness34.2%epoch.aiArchived copyAccuracyReported Evaluated Benchmark leaderboardEpoch AIIndependentRecord150004
GLM-5.2 (max)SWE-bench Verified — Mercor mini-swe-agent, 1000 steps78.5%mercor.comArchived copyResolved TasksReported Benchmark leaderboardMercorIndependentRecord150004
GLM-5.2 (max)GPQA Diamond — Mercor single-shot90.2%mercor.comArchived copyGPQA Diamond AccuracyReported Benchmark leaderboardMercorIndependentRecord150004
GLM-5.2SWE Atlas Codebase QnA — Public; mini-swe-agent48.1%labs.scale.comArchived copyMean pass@1Reported Benchmark leaderboardScale AIIndependentRecord150004
GLM-5.2 (max)Humanity's Last Exam Text only — Mercor single-shot31.7%mercor.comArchived copyHumanity's Last Exam AccuracyReported Benchmark leaderboardMercorIndependentRecord150004
GLM-5.2 (max)MMLU-Pro Original — Mercor single-shot85.9%mercor.comArchived copyAccuracyReported Benchmark leaderboardMercorIndependentRecord150004
GLM-5.2SWE Atlas Refactoring — mini-swe-agent42.4%labs.scale.comArchived copyMean pass@1Reported Benchmark leaderboardScale AIIndependentRecord150004
GLM-5.2 (max)BrowseComp 130-question subset — Mercor web research agent65.4%mercor.comArchived copyBrowseComp AccuracyReported Benchmark leaderboardMercorIndependentRecord150004
GLM-5.2 (low)GPQA Diamond — Epoch AI harness87.9%epoch.aiArchived copyGPQA Diamond AccuracyReported Evaluated Benchmark leaderboardEpoch AIIndependentRecord150004
GLM-5.2 (max)SWE-bench Verified — Epoch AI harness78.7%epoch.aiArchived copyResolved TasksReported Evaluated Benchmark leaderboardEpoch AIIndependentRecord150004
GLM-5.2 (max)SWE Atlas Codebase QnA — Mercor mini-swe-agent, 250 steps46.0%mercor.comArchived copyMean pass@1Reported Benchmark leaderboardMercorIndependentRecord150004
GLM-5.2 (none)GPQA Diamond — Epoch AI harness71.2%epoch.aiArchived copyGPQA Diamond AccuracyReported Evaluated Benchmark leaderboardEpoch AIIndependentRecord150004
GLM-5.2 (max)DeepSWE 1.1 — Mercor mini-swe-agent, 500 steps38.9%mercor.comArchived copyDeepSWE Pass@1Reported Benchmark leaderboardMercorIndependentRecord150004
GLM-5.2 (max)APEX-Agents Original45.2%mercor.comArchived copyPass rateReported Benchmark leaderboardMercorIndependentRecord150004
GLM-5.2 (max)GPQA Diamond — Epoch AI harness91.9%epoch.aiArchived copyGPQA Diamond AccuracyReported Evaluated Benchmark leaderboardEpoch AIIndependentRecord150004
GLM-5.3 (max)APEX-Agents Original56.6%mercor.comArchived copyPass rateReported Benchmark leaderboardMercorIndependentRecord150005
GLM-5.3 (max)SimpleQA Verified Original — Epoch AI harness41.0%epoch.aiArchived copyAccuracyReported Evaluated Benchmark leaderboardEpoch AIIndependentRecord150005
GLM-5.3 (low)CursorBench 4.033.3%cursor.comArchived copyCursorBench AccuracyReported Benchmark leaderboardCursorIndependentRecord150005
GLM-5.3 (max)SWE-bench Verified — Mercor mini-swe-agent, 1000 steps81.7%mercor.comArchived copyResolved TasksReported Benchmark leaderboardMercorIndependentRecord150005
GLM-5.3 (Max)Riemann-bench October 2026 leaderboard12.0%surgehq.aiArchived copyRiemann-bench Pass@1Reported Benchmark leaderboardSurge AIIndependentRecord150005
GLM-5.3FrontierSWE 2 — October 2026 evaluation30.2%frontierswe.comArchived copyFrontierSWE ScoreReported Benchmark leaderboardProximalIndependentRecord150005
GLM-5.3 (max)Terminal-Bench 4.0 — Mercor mini-swe-agent37.9%mercor.comArchived copyTerminal-Bench AccuracyReported Benchmark leaderboardMercorIndependentRecord150005
GLM-5.3 (max)CursorBench 4.042.6%cursor.comArchived copyCursorBench AccuracyReported Benchmark leaderboardCursorIndependentRecord150005
GLM-5.3 (max)Humanity's Last Exam Text only — Mercor single-shot40.0%mercor.comArchived copyHumanity's Last Exam AccuracyReported Benchmark leaderboardMercorIndependentRecord150005
GLM-5.3 (max)SWE Atlas Codebase QnA — Mercor mini-swe-agent, 250 steps42.5%mercor.comArchived copyMean pass@1Reported Benchmark leaderboardMercorIndependentRecord150005
GLM-5.3 (max)MMLU-Pro Original — Mercor single-shot88.0%mercor.comArchived copyAccuracyReported Benchmark leaderboardMercorIndependentRecord150005
GLM-5.3 (max)GPQA Diamond — Mercor single-shot91.3%mercor.comArchived copyGPQA Diamond AccuracyReported Benchmark leaderboardMercorIndependentRecord150005
GLM-5.3 (high)CursorBench 4.038.0%cursor.comArchived copyCursorBench AccuracyReported Benchmark leaderboardCursorIndependentRecord150005
GLM-5.3 (max)DeepSWE 1.1 — Mercor mini-swe-agent, 500 steps70.5%mercor.comArchived copyDeepSWE Pass@1Reported Benchmark leaderboardMercorIndependentRecord150005
GLM-5.3 (max)GPQA Diamond — Epoch AI harness90.9%epoch.aiArchived copyGPQA Diamond AccuracyReported Evaluated Benchmark leaderboardEpoch AIIndependentRecord150005
GLM-5.3 (max)SWE-bench Multilingual — Mercor mini-swe-agent, 1000 steps87.6%mercor.comArchived copyResolved TasksReported Benchmark leaderboardMercorIndependentRecord150005
GLM-5.3-Flash (max)SWE-bench Multilingual — Mercor mini-swe-agent, 1000 steps88.9%mercor.comArchived copyResolved TasksReported Benchmark leaderboardMercorIndependentRecord150006
GLM-5.3-Flash (max)Terminal-Bench 4.0 — Mercor mini-swe-agent29.8%mercor.comArchived copyTerminal-Bench AccuracyReported Benchmark leaderboardMercorIndependentRecord150006
GLM-5.3-FlashFrontierSWE 2 — October 2026 evaluation18.1%frontierswe.comArchived copyFrontierSWE ScoreReported Benchmark leaderboardProximalIndependentRecord150006

Z.ai models