Zhipu (GLM) · model
GLM-5.3-Flash benchmark scores
As of 2026-11-21, GLM-5.3-Flash (Zhipu (GLM)) has measured scores on 43 benchmarks and estimated scores on 113 more.
| Benchmark | Score | Source |
|---|---|---|
| AA-GPQA Diamond | 86.7% | estimated ± 3.6 pp, high confidence |
| AA-HLE | 29.9% | estimated ± 6.2 pp, medium confidence |
| AA-SciCode | 54.8% | estimated ± 2.5 pp, high confidence |
| AIME26 | 95.9% | estimated ± 1.2 pp, medium confidence |
| Artificial Analysis Intelligence Index | 25.7% | estimated ± 5.8 pp, medium confidence |
| CritPt | 5.9% | estimated ± 5.3 pp, medium confidence |
| GDPval-AA | 63.0% | estimated ± 5.6 pp, medium confidence |
| GPQA Diamond | 88.8% | estimated ± 1.8 pp, high confidence |
| HLE | 40.1% | estimated ± 9.2 pp, low confidence |
| LiveCodeBench v6 | 94.0% | estimated ± 1.9 pp, medium confidence |
| AA-Omniscience Accuracy | 36.7% | estimated ± 6.5 pp, medium confidence |
| SciCode | 37.8% | estimated ± 2.9 pp, low confidence |
| Terminal-Bench 2.1 | 84.3% | measured |
| BrowseComp | 89.0% | estimated ± 4.0 pp, high confidence |
| HLE w/ tools | 55.3% | measured |
| τ²-bench results | 97.9% | estimated ± 3.1 pp, high confidence |
| AA Coding Index | 73.1% | estimated ± 1.2 pp, high confidence |
| AA-MMMU-Pro | 79.1% | estimated ± 2.3 pp, high confidence |
| APEX-Agents | 37.6% | estimated ± 7.4 pp, low confidence |
| SWE-bench Verified | 81.6% | estimated ± 2.3 pp, high confidence |
| AutomationBench | 48.8% | measured |
| CyberGym | 84.1% | estimated ± 3.1 pp, low confidence |
| SWE-bench Pro | 62.1% | estimated ± 3.5 pp, high confidence |
| WideResearch | 85.9% | estimated ± 3.9 pp, low confidence |
| AA Agentic Index | 56.2% | estimated ± 3.3 pp, medium confidence |
| MMLU-Pro | 85.5% | estimated ± 4.8 pp, high confidence |
| FailureSensorIQ | 64.9% | estimated ± 2.1 pp, high confidence |
| GPQA | 88.1% | estimated ± 4.5 pp, high confidence |
| MMLU | 93.4% | estimated ± 1.2 pp, medium confidence |
| Vals GPQA Diamond | 86.4% | measured |
| Vals LiveCodeBench | 80.5% | measured |
| Vals MMLU-Pro | 86.1% | measured |
| MMMU | 85.1% | estimated ± 2.9 pp, low confidence |
| Chess Puzzles | 21.0% | measured |
| AA MMLU-Pro | 87.2% | estimated ± 1.1 pp, high confidence |
| Gert Labs | 77.5% | estimated ± 9.1 pp, low confidence |
| AA-AnalystAgent | 46.1% | estimated ± 5.8 pp, low confidence |
| AA EnterpriseOps-Gym | 47.0% | estimated ± 7.2 pp, low confidence |
| AA Harvey LAB v1.0 | 89.6% | estimated ± 1.7 pp, medium confidence |
| AA Tau3 Banking | 47.2% | measured |
| AA Terminal-Bench 2.1 | 84.3% | measured |
| ApprenticeBench | 13.1% | estimated ± 12.5 pp, low confidence |
| ARC-AGI-1 | 80.7% | estimated ± 6.5 pp, medium confidence |
| ARC-AGI-2 | 43.6% | estimated ± 13.8 pp, low confidence |
| BabaIsBench | 67.0% | measured |
| CursorBench 3.1 | 66.4% | estimated ± 5.8 pp, low confidence |
| CursorBench 3.2 | 63.7% | estimated ± 2.0 pp, high confidence |
| EBR-bench | 0.0% | estimated ± 10.4 pp, low confidence |
| Terminal-Bench 3.0 | 29.5% | estimated ± 3.0 pp, high confidence |
| FrontierCode 1.1 Main | 51.1% | estimated ± 3.0 pp, medium confidence |
| FrontierMath v2 (Tier 4) | 17.1% | measured |
| FrontierMath v2 (Tiers 1-3) | 55.8% | measured |
| FrontierSWE v2 | 18.1% | measured |
| Furniture Assembly | 32.9% | estimated ± 9.3 pp, medium confidence |
| MaxxBench | 55.1% | measured |
| MazeBench | 0.0% | measured |
| Mystery Game Puzzles | 8.0% | measured |
| NYT Connections | 77.8% | estimated ± 6.4 pp, low confidence |
| OfficeQA Pro | 62.4% | measured |
| OSWorld-Verified | 83.9% | estimated ± 9.0 pp, low confidence |
| ParseBench-1 | 69.4% | measured |
| Terminal-Bench 4.1 | 37.0% | measured |
| Terminal-Bench Hard | 62.7% | estimated ± 2.8 pp, medium confidence |
| Vals SWE-bench | 92.0% | measured |
| Terminal-Bench 2.1 (Vals AI) | 62.9% | measured |
| VulcanBench v3 | 78.2% | estimated ± 3.9 pp, medium confidence |
| CWE-Bench-AA | 55.8% | measured |
| AA Cyber Index | 49.9% | measured |
| DeepsecBench-AA | 19.9% | measured |
| AA ITBench | 51.2% | measured |
| BenchCAD Vision2Code (no tools) | 0.3% | estimated ± 0.1 pp, low confidence |
| BenchCAD Vision2Code (tools) | 0.7% | estimated ± 0.0 pp, low confidence |
| Bug Hunt Bench | 17.7% | measured |
| Chartography (no tools) | 27.3% | estimated ± 6.0 pp, low confidence |
| Chartography (tools) | 78.0% | measured |
| CursorBench 4.0 | 36.8% | measured |
| CWE-bench v1 | 57.4% | estimated ± 5.1 pp, medium confidence |
| DeepSWE | 63.4% | measured |
| DRACO | 86.9% | estimated ± 0.7 pp, low confidence |
| FrontierCode 2.0 | 27.1% | estimated ± 4.7 pp, medium confidence |
| FrontierPhysics | 2.5% | estimated ± 5.0 pp, low confidence |
| HLE w/o tools | 34.1% | estimated ± 4.9 pp, high confidence |
| OfficeQA | 74.7% | estimated ± 0.2 pp, low confidence |
| OSWorld 2.0 | 59.0% | estimated ± 5.5 pp, low confidence |
| OSWorld 2.1 | 66.4% | estimated ± 6.1 pp, medium confidence |
| OSWorld 2.1 Offline | 63.2% | estimated ± 7.4 pp, low confidence |
| PostTrainBench v1.1 | 35.6% | estimated ± 1.7 pp, low confidence |
| SpireClimb | 0.0% | estimated ± 9.3 pp, medium confidence |
| StealthMark | 47.9% | estimated ± 7.7 pp, medium confidence |
| SWE-bench Multimodal | 56.6% | estimated ± 5.6 pp, low confidence |
| Terminal-Bench 4.0 | 41.3% | estimated ± 4.5 pp, high confidence |
| Terminal-Bench-Science 0.1 | 20.8% | estimated ± 10.7 pp, low confidence |
| Toolathlon-Verified | 78.4% | measured |
| VulcanBench Frontier v4 | 85.0% | estimated ± 1.8 pp, high confidence |
| AA AutomationBench | 54.6% | estimated ± 5.5 pp, medium confidence |
| GDP.pdf | 19.9% | estimated ± 3.1 pp, high confidence |
| AA Terminal-Bench 4.0 | 36.6% | estimated ± 6.2 pp, medium confidence |
| ARC-AGI-3 | 0.0% | estimated ± 8.6 pp, low confidence |
| cursorBench41 | 37.9% | estimated ± 2.6 pp, medium confidence |
| CWE-bench | 44.1% | estimated ± 4.3 pp, low confidence |
| MKA | 36.2% | measured |
| SkillsBench official leaderboard (BenchFlow) | 49.4% | estimated ± 6.2 pp, medium confidence |
| CompileBench | 57.8% | estimated ± 10.0 pp, low confidence |
| FACTS Multimodal | 36.6% | estimated ± 5.6 pp, medium confidence |
| Hermes Bench | 68.9% | measured |
| Hermes Index | 34.9% | measured |
| ParseBench-2 | 74.9% | estimated ± 4.3 pp, high confidence |
| CharXiv | 89.4% | measured |
| CharXiv w/o tools | 84.5% | estimated ± 1.8 pp, medium confidence |
| ExploitBench | 16.5% | estimated ± 7.2 pp, low confidence |
| ExploitGym | 11.9% | estimated ± 6.3 pp, medium confidence |
| AI2D_TEST | 93.5% | estimated ± 0.9 pp, low confidence |
| CountBench | 98.3% | estimated ± 3.7 pp, low confidence |
| DynaMath | 87.9% | estimated ± 0.8 pp, low confidence |
| ERQA | 70.8% | estimated ± 5.6 pp, medium confidence |
| HMMT Feb 2026 | 89.9% | estimated ± 1.5 pp, medium confidence |
| HMMT Nov 2025 | 94.8% | estimated ± 1.5 pp, medium confidence |
| MathVision | 88.6% | estimated ± 0.9 pp, high confidence |
| MCP Atlas | 82.5% | estimated ± 6.0 pp, low confidence |
| MLVU (M-Avg) | 88.8% | estimated ± 1.5 pp, high confidence |
| MMAnswerBench | 88.0% | estimated ± 2.4 pp, medium confidence |
| MMMU-Pro | 77.9% | estimated ± 2.5 pp, medium confidence |
| NL2Repo | 56.3% | measured |
| PostTrainBench v1 | 40.4% | estimated ± 1.3 pp, low confidence |
| RealWorldQA | 87.1% | estimated ± 1.5 pp, high confidence |
| ScreenSpot Pro | 84.5% | estimated ± 6.8 pp, medium confidence |
| Terminal-Bench 2.0 | 68.3% | estimated ± 5.8 pp, medium confidence |
| Toolathlon | 55.9% | estimated ± 5.1 pp, low confidence |
| Video-MME (with subtitle) | 90.5% | estimated ± 2.9 pp, high confidence |
| VideoMMMU | 87.1% | estimated ± 1.5 pp, medium confidence |
| We-Math | 84.6% | estimated ± 7.9 pp, low confidence |
| AA LiveCodeBench | 99.0% | estimated ± 2.6 pp, medium confidence |
| MedXpertQA (MM) | 78.8% | estimated ± 6.4 pp, low confidence |
| React Native Evals | 86.7% | estimated ± 4.7 pp, low confidence |
| PostTrainBench | 36.8% | measured |
| DAYJOB | 1.3% | estimated ± 3.2 pp, high confidence |
| DAYJOB Healthcare | 0.7% | estimated ± 4.2 pp, high confidence |
| Finance Agent v2 | 56.1% | estimated ± 3.8 pp, medium confidence |
| FrontierCode 1.1 Extended | 52.7% | estimated ± 2.4 pp, low confidence |
| HLE-Verified | 3.2% | estimated ± 4.6 pp, low confidence |
| LABBench2 | 80.9% | estimated ± 1.8 pp, low confidence |
| ScarfBench | 64.0% | estimated ± 3.8 pp, medium confidence |
| BlenderWorld (text-only) | 71.4% | measured |
| Cookie-Bench | 76.9% | measured |
| Agents' Last Exam | 26.3% | measured |
| OpenHarmony Bench v1.0 | 57.3% | measured |
| OpenHarmony Bench v1.2 | 50.4% | measured |
| SEC-Bench Pro | 34.4% | estimated ± 9.3 pp, medium confidence |
| BabyVision | 53.4% | measured |
| MMVU | 80.5% | measured |
| SWE-Marathon | 41.1% | estimated ± 2.1 pp, low confidence |
| LifeVal | 75.2% | estimated ± 3.8 pp, medium confidence |
| MMMU-Pro w/ Python | 83.0% | estimated ± 2.7 pp, medium confidence |
| ExploitBench Internal Port | 0.0% | estimated ± 1.4 pp, low confidence |
| MathVision w/ Python | 95.4% | estimated ± 1.0 pp, low confidence |
| SWE-Atlas Codebase QnA | 58.2% | estimated ± 4.8 pp, high confidence |