GLM-5.1 places in the top three on 11 benchmarks.
Open weights Flagship
Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| LMArena (Elo) | 1,465 | #46 of 335 | Official leaderboard | |
| Structured Output Benchmark | 86.6% | #3 of 19 | Official leaderboard | |
| EQ-Bench 3 | 1,539 | #13 of 78 | Official leaderboard | |
| Humanity's Last Exam | 52.3% | #34 of 154 | Lab model card | |
| Longform Creative Writing (EQ-Bench) | 73.5 | #31 of 119 | Official leaderboard | |
| Creative Writing v3 | 72.2 | #44 of 120 | Official leaderboard | |
| EvoArena SWE-Chain-Evo | 36.1% | #3 of 8 | Official leaderboard | |
| EvoArena Terminal-Bench-Evo | 55.3% | #3 of 8 | Official leaderboard | |
| Judgemark v4 | 67.2 | #18 of 40 | Official leaderboard | |
| Extended NYT Connections | 77.7% | #48 of 103 | Official leaderboard | |
| CompassArena | 1,006 | #5 of 10 | Official leaderboard | |
| EvoArena PersonaMem-Evo | 50.4% | #4 of 8 | Official leaderboard | |
| SEA-HELM | 66.9 | #38 of 71 | Official leaderboard | |
| SimpleQA Verified | 34% | #64 of 94 | Epoch AI run | |
| BullshitBench V2 | 22% | #88 of 129 | Official leaderboard | |
| DesignArena ASCII Art | 1,141 | #53 of 77 | Official leaderboard | |
| Lech Mazur Creative Writing | -1.08 | #35 of 50 | Official leaderboard | |
| AA-LCR Extended | 37.9% | #42 of 49 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| GPQA Diamond | 89.9% | #49 of 229 | Epoch AI run | |
| FrontierMath | 33.5% | #13 of 55 | Official leaderboard | |
| OTIS Mock AIME 2024-2025 | 93.3% | #58 of 212 | Epoch AI run | |
| HMMT Nov 2025 | 94% | #8 of 28 | Lab model card | |
| Echo General AI Prediction Leaderboard | 1,005 | #5 of 17 | Official leaderboard | |
| MathArena Apex Shortlist | 71.8% | #13 of 38 | Official leaderboard | |
| FrontierMath Tier 4 | 12.5% | #19 of 55 | Epoch AI run | |
| MathArena (Apex, IMO, USAMO) | 11.5% | #20 of 47 | Official leaderboard | |
| AIME 2026 | 95.8% | #14 of 31 | Official leaderboard | |
| ProofBench | 22.2% | #28 of 59 | Official leaderboard | |
| HMMT 2026 | 89.4% | #15 of 31 | Official leaderboard | |
| IMO-AnswerBench | 83.8% | #18 of 37 | Lab model card | |
| SimpleBench | 55.1% | #52 of 106 | Official leaderboard | |
| ResearchClawBench | 18.2 | #12 of 23 | Official leaderboard | |
| FrontierMath Tiers 1-3 v2 | 36.8% | #60 of 86 | Official leaderboard | |
| NatureBench | 6.67% | #15 of 18 | Official leaderboard | |
| HorizonMath | 1.80% | #7 of 8 | Official leaderboard | |
| Project Euler (MathArena) | 67.2% | #4 of 4 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| DesignArena Website | 1,263 | #42 of 151 | Official leaderboard | |
| QwenSVG | 1,605 | #2 of 7 | Official leaderboard | |
| DesignArena Game Dev | 1,256 | #41 of 143 | Official leaderboard | |
| DesignArena Data Visualization | 1,258 | #36 of 118 | Official leaderboard | |
| LiveSQLBench | 29% | #11 of 36 | Official leaderboard | |
| EvoCode-Bench | 36.2% | #4 of 13 | Official leaderboard | |
| DesignArena Code | 1,261 | #45 of 143 | Official leaderboard | |
| DesignArena SVG | 1,209 | #33 of 103 | Official leaderboard | |
| DesignArena UI Component | 1,260 | #45 of 139 | Official leaderboard | |
| WeirdML | 57.1% | #37 of 112 | Official leaderboard | |
| DesignArena 3D Design | 1,234 | #46 of 136 | Official leaderboard | |
| OpenHands Index | 58.2% | #12 of 33 | Official leaderboard | |
| Claw-SWE-Bench | 73.1% | #6 of 16 | Official leaderboard | |
| ALE-Bench | 1,646 | #50 of 126 | Official leaderboard | |
| WebDev Arena | 1,508 | #43 of 105 | arena.ai, via Epoch AI | |
| SWE-Bench Pro | 58.4% | #34 of 81 | Lab model card | |
| RoadmapBench | 18.3% | #5 of 11 | Official leaderboard | |
| Code Arena | 1,508 | #59 of 128 | Official leaderboard | |
| SWE-bench Verified | 74.2% | #16 of 32 | Epoch AI run | |
| DesignArena Android (Kotlin) | 1,173 | #24 of 46 | Official leaderboard | |
| DesignArena Agentic Overall | 1,154 | #42 of 74 | Official leaderboard | |
| DesignArena Full Stack | 1,145 | #28 of 49 | Official leaderboard | |
| Code Migration | 25.8% | #47 of 78 | Official leaderboard | |
| DesignArena React Native | 1,154 | #34 of 56 | Official leaderboard | |
| FeatureBench | 44.2% | #10 of 16 | Official leaderboard | |
| FeatureBench Lite | 44.2% | #10 of 16 | Official leaderboard | |
| Vibe Code Bench | 31.5% | #78 of 120 | Official leaderboard | |
| AndroidBench | 63.2% | #21 of 32 | Official leaderboard | |
| DesignArena Frontend Web App | 1,158 | #35 of 53 | Official leaderboard | |
| NL2Repo-Bench | 42.7% | #20 of 30 | Lab model card | |
| ContextBench | 23.6% | #6 of 9 | Official leaderboard | |
| QwenWebDev | 1,564 | #4 of 6 | Official leaderboard | |
| FrogsGame post-training task | 0.40% | #12 of 17 | Official leaderboard | |
| APEX-SWE | 34.5% | #46 of 65 | Official leaderboard | |
| DesignArena Agentic Game Dev | 1,111 | #20 of 27 | Official leaderboard | |
| ParallelKernelBench | 6.90% | #3 of 4 | Official leaderboard | |
| FrontierSWE v1 | 25.7% | #13 of 17 | Official leaderboard | |
| SWE-Marathon | 1.00% | #10 of 12 | Official leaderboard | |
| SlopCodeBench | 13.8% | #19 of 22 | Official leaderboard | |
| GameCraft-Bench | 18.3% | #13 of 15 | Official leaderboard | |
| DesignArena Godot Game Dev | 1,039 | #22 of 24 | Official leaderboard | |
| EvoCode-Bench v2 | 5.90% | #12 of 13 | Official leaderboard | |
| SWE-Together | 52% | #21 of 22 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| PinchBench | 80.9% | #2 of 48 | Official leaderboard | |
| LiveClawBench | 74.7% | #2 of 17 | Official leaderboard | |
| Terminal-Bench 2.0 | 69% | #8 of 63 | Lab model card | |
| SkillsBench | 58.4% | #4 of 22 | Official leaderboard | |
| TERMS-Bench | 0.69 | #2 of 11 | Official leaderboard | |
| Claw-Eval Pass^3 | 62.3% | #6 of 24 | Official leaderboard | |
| QwenClawBench | 58.7% | #3 of 12 | Official leaderboard | |
| ClawProBench | 62.9% | #14 of 49 | Official leaderboard | |
| VitaBench 2.0 | 42.3% | #5 of 17 | Official leaderboard | |
| GAIA | 68.8% | #20 of 64 | Official leaderboard | |
| PACT | 1,554 | #9 of 26 | Official leaderboard | |
| SnakeBench | 26 | #67 of 188 | Official leaderboard | |
| BrowseComp | 79.3% | #29 of 81 | Lab model card | |
| Toolathlon | 40.7% | #18 of 50 | Official leaderboard | |
| Clocktower Radio | 1,655 | #9 of 23 | Official leaderboard | |
| Claw-Eval | 77.3% | #10 of 24 | Official leaderboard | |
| TUA-Bench | 48.1% | #5 of 12 | Official leaderboard | |
| tau3-bench | 70.6% | #6 of 14 | Lab model card | |
| ClawBench V2 | 34.6% | #3 of 7 | Official leaderboard | |
| WildClawBench | 48.2% | #14 of 31 | Official leaderboard | |
| MCP Atlas | 75.6% | #21 of 37 | Official leaderboard | |
| CivBench | 1,482 | #12 of 19 | Official leaderboard | |
| DeepTerminalBench | 30% | #6 of 9 | Official leaderboard | |
| PinchBench 2.0 | 60% | #42 of 57 | Official leaderboard | |
| AutomationBench | 6.20% | #31 of 42 | Official leaderboard | |
| Chess Puzzles | 20% | #71 of 93 | Epoch AI run | |
| EdgeBench | 37.4 | #4 of 5 | Official leaderboard | |
| Gaia2-CLI | 50.5% | #6 of 7 | Official leaderboard | |
| Long-Horizon-Terminal-Bench (LHTB) | 26.7% | #22 of 25 | Official leaderboard | |
| CEO-Bench | $0 | #18 of 19 | Official leaderboard | |
| Terminal-Bench 2.1 | 58.6% | #13 of 13 | Official leaderboard | |
| ToolcallFormatIFBench | 60.1% | #5 of 5 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Vals Finance Agent | 57.7% | #10 of 49 | Official leaderboard | |
| Workspace-Bench-Lite | 61% | #2 of 9 | Official leaderboard | |
| FORTE | 73.8% | #5 of 21 | Official leaderboard | |
| LegalBench | 84.4% | #33 of 138 | Official leaderboard | |
| Vals Index v1 | 57.9% | #10 of 34 | Official leaderboard | |
| CorpFin v2 | 64.5% | #38 of 119 | Official leaderboard | |
| YC-Bench | $1,510,772 | #19 of 59 | Official leaderboard | |
| DesignArena PDF | 1,200 | #15 of 37 | Official leaderboard | |
| Vending-Bench 2 | $5,634 | #29 of 67 | Official leaderboard | |
| CHI-Bench | 18.7% | #11 of 23 | Official leaderboard | |
| MedCode | 41.6% | #50 of 96 | Official leaderboard | |
| Public Benefits Bench | 61.8% | #29 of 50 | Official leaderboard | |
| DesignArena Python-PPTX Slides | 1,150 | #21 of 36 | Official leaderboard | |
| TaxEval v2 | 71.2% | #76 of 129 | Official leaderboard | |
| E-Commerce Bench | 2.26 | #12 of 19 | Official leaderboard | |
| DesignArena HTML Slides | 1,066 | #18 of 28 | Official leaderboard | |
| Legal Research Bench | 27.9% | #63 of 93 | Official leaderboard | |
| Vals Finance Agent v2 | 44.8% | #57 of 80 | Official leaderboard | |
| APEX-Agents | 40.9% | #48 of 61 | Collected by Epoch AI, source not named | |
| APEX-Agents 1.1 | 56.8% | #52 of 65 | Official leaderboard | |
| Agents' Last Exam | 11.5% | #25 of 31 | Official leaderboard | |
| MedScribe | 72.3% | #80 of 97 | Official leaderboard | |
| ProfBench | 41.6% | #60 of 72 | Official leaderboard | |
| CaseLaw v2 | 51.5% | #48 of 53 | Official leaderboard | |
| Harvey LAB (Legal Agent Benchmark) | 0% | #73 of 80 | Official leaderboard | |
| APEX-1 | 52.4% | #66 of 71 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| MineBench | 1,501 | #39 of 75 | Official leaderboard | |
| VoxelBench | 1,443 | #42 of 56 | Official leaderboard | |
| AgentWorldBench | 51.3 | #10 of 13 | Official leaderboard | |
| QwenWorldBench | 50.2% | #4 of 5 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| CyberGym | 84% | #9 of 34 | Official leaderboard | |
| nocot-bench (NCRI) | 75.2 | #74 of 185 | Official leaderboard | |
| CRSBench | 68.6% | #2 of 5 | Official leaderboard | |
| BioTIER Permit | 99.4% | #35 of 69 | Official leaderboard | |
| Human Pathogen Capabilities Test v2 (HPCT v2) | 50.3% | #43 of 71 | Official leaderboard | |
| BiomniBench | 60.4 | #5 of 8 | Official leaderboard | |
| ExploitBench | 18% | #6 of 9 | Official leaderboard | |
| HelpBench | 82% | #6 of 9 | Official leaderboard | |
| BioTIER Refuse | 23.7% | #52 of 69 | Official leaderboard | |
| EvalDetectBench | 0.78 | #9 of 10 | Official leaderboard | |
| ExploitGym | 0.46% | #18 of 18 | Official leaderboard | |
| SecCodeBench | 49% | #6 of 6 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| CAIS Text Capabilities Index | 27.6 | #39 of 68 | Official leaderboard | |
| CAIS Risk Index | 58.0 | #46 of 67 | Official leaderboard |
The Self-Reported Index fits one score to the 13 capability results Z.ai reported for this model, on the scale of the Epoch Capabilities Index. Labs choose which benchmarks to show and how to run them, so read it as the lab's claim, not an independent measurement. The 90% interval runs from 151.3 to 155.9. How we compute it.
Where the benchmark's own leaderboard, or Epoch AI's run, also has this model. Gaps mostly come from different harnesses, effort levels and task subsets. Every pair.
| Benchmark | Lab says | Board says | Gap | What differs |
|---|---|---|---|---|
| CyberGym | 68.7% CyberGym | 84% Leaderboard, dynamic, Xuanwu Atuin AI | −15 pts | The lab uses Claude Code, while the board pairs GLM-5.1 with Tencent's Xuanwu Atuin AI agent. |
| HMMT 2026 | 82.6% HMMT Feb. 2026 | 89.4% Leaderboard | −6.8 pts | |
| MCP Atlas | 71.8% MCP-Atlas (Public Set) | 75.6% Leaderboard | −3.8 pts | |
| GPQA Diamond | 86.2% GPQA-Diamond | 89.9% Epoch AI run | −3.7 pts | |
| AIME 2026 | 95.3% AIME 2026 | 95.8% Leaderboard | −0.5 pts | |
| Toolathlon | 40.7% Tool-Decathlon | 40.7% Leaderboard | Same | |
| Vending-Bench 2 | $5,634 Vending Bench 2 | $5,634 Leaderboard | Same |
Z.ai's models in color, against every other model on the Epoch Capabilities Index, by release date.
Source: Epoch Capabilities Index, Epoch AI (CC BY 4.0)
The benchmarks the lab chose to headline, with the scores it gave.