GLM-4.7 places in the top three on 5 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) | 1,441 | #77 of 335 | Official leaderboard | |
| MMLU-Pro | 84.3% | #40 of 167 | Lab model card | |
| AMemGym | 47.8% | #3 of 11 | Official leaderboard | |
| Structured Output Benchmark | 86.1% | #6 of 19 | Official leaderboard | |
| EQ-Bench 3 | 1,463 | #26 of 78 | Official leaderboard | |
| Humanity's Last Exam | 42.8% | #60 of 154 | Lab model card | |
| Judgemark v2.1 | 77.5 | #17 of 42 | Official leaderboard | |
| SEA-HELM | 71.5 | #30 of 71 | Official leaderboard | |
| Longform Creative Writing (EQ-Bench) | 66.0 | #52 of 119 | Official leaderboard | |
| HUMAINE | 29 | #22 of 50 | Official leaderboard | |
| Creative Writing v3 | 70.2 | #54 of 120 | Official leaderboard | |
| DesignArena ASCII Art | 1,160 | #40 of 77 | Official leaderboard | |
| StudyArena Text | 1,221 | #54 of 85 | Official leaderboard | |
| LOCA-Bench | 5.30% | #6 of 9 | Official leaderboard | |
| CL-bench | 15.9% | #17 of 24 | Official leaderboard | |
| Vectara Hallucination Leaderboard (2025 dataset) | 11.7% | #75 of 104 | Official leaderboard | |
| SimpleQA Verified | 32.2% | #68 of 94 | Epoch AI run | |
| CL-bench Life | 10.9% | #12 of 16 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| AIME 2025 | 95.7% | #22 of 143 | Lab model card | |
| HMMT 2025 | 97.1% | #9 of 54 | Lab model card | |
| HMMT Nov 2025 | 93.5% | #9 of 28 | Lab model card | |
| General365 | 57.4% | #8 of 23 | Official leaderboard | |
| GPQA Diamond | 83.3% | #87 of 229 | Epoch AI run | |
| OTIS Mock AIME 2024-2025 | 83.3% | #92 of 212 | Epoch AI run | |
| IMO-AnswerBench | 82% | #20 of 37 | Lab model card | |
| SimpleBench | 47.7% | #62 of 106 | Official leaderboard | |
| ProofBench | 6.00% | #49 of 59 | Official leaderboard | |
| FrontierMath | 2.44% | #47 of 55 | Official leaderboard | |
| FrontierMath Tier 4 | 0% | #53 of 55 | Epoch AI run |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| LiveCodeBench v6 | 84.9% | #21 of 90 | Lab model card | |
| SWE-bench Verified | 73.8% | #54 of 155 | Lab model card | |
| DataPrep-Bench Math | 25.3 | #2 of 5 | Official leaderboard | |
| Bird-SQL | 62.9% | #6 of 13 | Official leaderboard | |
| DesignArena Website | 1,219 | #72 of 151 | Official leaderboard | |
| CF-Div2-Stepfun | 74.1% | #4 of 8 | Official leaderboard | |
| DesignArena Code | 1,207 | #72 of 143 | Official leaderboard | |
| DesignArena 3D Design | 1,173 | #69 of 136 | Official leaderboard | |
| DesignArena Game Dev | 1,178 | #75 of 143 | Official leaderboard | |
| DesignArena UI Component | 1,190 | #75 of 139 | Official leaderboard | |
| WebDev Arena | 1,435 | #60 of 105 | arena.ai, via Epoch AI | |
| OTelBench | 13% | #8 of 14 | Official leaderboard | |
| DesignArena Data Visualization | 1,193 | #70 of 118 | Official leaderboard | |
| DataPrep-Bench Finance | 53.3 | #3 of 5 | Official leaderboard | |
| DataPrep-Bench Science | 23.5 | #3 of 5 | Official leaderboard | |
| Code Arena | 1,435 | #77 of 128 | Official leaderboard | |
| DesignArena SVG | 1,133 | #64 of 103 | Official leaderboard | |
| IOI (Vals) | 7.58% | #35 of 55 | Official leaderboard | |
| LiveSQLBench | 20.3% | #25 of 36 | Official leaderboard | |
| SWE-bench Multilingual | 66.7% | #51 of 73 | Lab model card | |
| ABC-Bench | 40.1% | #5 of 7 | Official leaderboard | |
| PRDBench | 38.4% | #3 of 4 | Official leaderboard | |
| DesignArena Android (Kotlin) | 1,097 | #36 of 46 | Official leaderboard | |
| DesignArena React Native | 1,099 | #44 of 56 | Official leaderboard | |
| OpenHands Index | 42.3% | #26 of 33 | Official leaderboard | |
| DataPrep-Bench Medical | 34.2 | #4 of 5 | Official leaderboard | |
| DesignArena Agentic Overall | 1,055 | #60 of 74 | Official leaderboard | |
| FeatureBench | 41.2% | #13 of 16 | Official leaderboard | |
| FeatureBench Lite | 41.2% | #13 of 16 | Official leaderboard | |
| ALE-Bench | 1,011 | #107 of 126 | Official leaderboard | |
| DesignArena Full Stack | 1,028 | #42 of 49 | Official leaderboard | |
| OJBench | 50.8% | #7 of 8 | Official leaderboard | |
| PostTrainBench | 7.48% | #21 of 23 | Official leaderboard | |
| DataPrep-Bench General | 77.3 | #5 of 5 | Official leaderboard | |
| DataPrep-Bench Law | 73.1 | #5 of 5 | Official leaderboard | |
| CC-Bench-V2 Backend | 19.6% | #3 of 3 | Official leaderboard | |
| CC-Bench-V2 Large Repo Exploration | 47.8% | #3 of 3 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Terminal-Bench Hard | 33.3% | #2 of 13 | Lab model card | |
| tau2-bench | 87.4% | #20 of 79 | Lab model card | |
| SnakeBench | 26 | #70 of 188 | Official leaderboard | |
| BrowseComp-ZH | 66.6% | #9 of 19 | Lab model card | |
| OpenThoughts-TBLite | 67.7% | #2 of 4 | Official leaderboard | |
| BrowseComp | 67.5% | #50 of 81 | Lab model card | |
| VITA-Bench | 18.3% | #12 of 18 | Official leaderboard | |
| LisanBench | 197.0 | #90 of 133 | Official leaderboard | |
| Chess Puzzles | 21% | #64 of 93 | Epoch AI run | |
| Terminal-Bench 2.0 | 33.4% | #30 of 43 | Official leaderboard | |
| Toolathlon | 23.8% | #37 of 50 | Official leaderboard | |
| ClawProBench | 54.6% | #38 of 49 | Official leaderboard | |
| Pencil Puzzle Bench | 0% | #44 of 53 | Official leaderboard | |
| RWSearch | 69% | #6 of 7 | Official leaderboard | |
| MCP Atlas | 58.1% | #32 of 37 | Official leaderboard | |
| BalatroBench | 3.07 | #15 of 16 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Vals Finance Agent v1 | 41.9% | #17 of 47 | Official leaderboard | |
| LegalBench | 83.4% | #55 of 138 | Official leaderboard | |
| ProfBench | 46.6% | #39 of 72 | Official leaderboard | |
| CaseLaw v2 | 54.9% | #37 of 53 | Official leaderboard | |
| TaxEval v2 | 68.8% | #91 of 129 | Official leaderboard | |
| Vending-Bench 2 | $2,377 | #49 of 67 | Official leaderboard | |
| Vals Finance Agent | 46% | #36 of 49 | Official leaderboard | |
| MedCode | 32.8% | #80 of 96 | Official leaderboard | |
| CorpFin v2 | 46.4% | #101 of 119 | Official leaderboard | |
| MedScribe | 68.6% | #87 of 97 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| MineBench | 1,100 | #62 of 75 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| BinaryAudit | 24.6% | #9 of 20 | Official leaderboard | |
| nocot-bench (NCRI) | 71.3 | #91 of 185 | Official leaderboard | |
| KORA | 25.2% | #30 of 42 | Official leaderboard | |
| CyberGym | 23.5% | #25 of 34 | Official leaderboard |
The Self-Reported Index fits one score to the 15 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 146.1 to 150.0. 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 |
|---|---|---|---|
| Terminal-Bench 2.0 | 41% Terminal Bench 2.0 | 33.4% Leaderboard, Terminus 2 | +7.6 pts |
| GPQA Diamond | 85.7% GPQA-Diamond | 83.3% Epoch AI run | +2.4 pts |
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.