GLM-5V-Turbo places in the top three on 3 benchmarks.
Closed 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 | |
|---|---|---|---|---|
| SimpleVQA | 78.2% | #3 of 30 | Lab model card | |
| LMArena (Elo) | 1,434 | #86 of 335 | Official leaderboard | |
| FACTS Benchmark Suite | 58.6% | #3 of 4 | Lab model card | |
| DesignArena ASCII Art | 1,113 | #62 of 77 | Official leaderboard | |
| Document Arena | 1,406 | #39 of 39 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Flame-React-Eval | 93.8% | #1 of 4 | Lab model card | |
| DesignArena Game Dev | 1,207 | #60 of 143 | Official leaderboard | |
| DesignArena 3D Design | 1,192 | #62 of 136 | Official leaderboard | |
| DesignArena Website | 1,224 | #70 of 151 | Official leaderboard | |
| DesignArena Code | 1,219 | #70 of 143 | Official leaderboard | |
| CC-Bench-V2 Large Repo Exploration | 72.2% | #1 of 2 | Lab model card | |
| DesignArena UI Component | 1,202 | #71 of 139 | Official leaderboard | |
| DesignArena Data Visualization | 1,200 | #68 of 118 | Official leaderboard | |
| DesignArena SVG | 1,140 | #61 of 103 | Official leaderboard | |
| DesignArena Full Stack | 1,128 | #32 of 49 | Official leaderboard | |
| Code Arena | 1,399 | #86 of 128 | Official leaderboard | |
| DesignArena Agentic Overall | 1,103 | #51 of 74 | Official leaderboard | |
| DesignArena React Native | 1,129 | #42 of 56 | Official leaderboard | |
| DesignArena Frontend Web App | 1,106 | #42 of 53 | Official leaderboard | |
| DesignArena Image to Frontend | 1,109 | #27 of 31 | Official leaderboard | |
| Image-to-WebDev Arena | 1,427 | #48 of 55 | Official leaderboard | |
| DesignArena Agentic Game Dev | 1,050 | #25 of 27 | Official leaderboard | |
| DesignArena Video to Website | 1,141 | #18 of 19 | Official leaderboard | |
| DesignArena Godot Game Dev | 940 | #23 of 24 | Official leaderboard | |
| DesignArena Image to HTML | 1,110 | #49 of 51 | Official leaderboard | |
| DesignArena Agentic Image to HTML | 1,105 | #11 of 11 | Official leaderboard | |
| Vision2Web | 31% | #7 of 7 | Lab model card | |
| CC-Bench-V2 Backend | 22.8% | #2 of 2 | Lab model card | |
| CC-Bench-V2 Frontend | 68.4% | #2 of 2 | Lab model card |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| OSWorld | 62.3% | #2 of 13 | Lab model card | |
| WebVoyager | 88.5% | #3 of 12 | Lab model card | |
| AndroidWorld | 75.7% | #6 of 23 | Lab model card | |
| SnakeBench | 28 | #50 of 188 | Official leaderboard | |
| PinchBench 2.0 | 67.6% | #36 of 57 | Official leaderboard | |
| BrowseComp-VL | 51.9% | #2 of 3 | Lab model card | |
| GameHorizon | 63.2 | #27 of 37 | Official leaderboard | |
| Claw-Eval Pass^3 | 53.8% | #18 of 24 | Official leaderboard | |
| Claw-Eval | 74% | #19 of 24 | Official leaderboard | |
| PinchBench | 80.7% | #13 of 14 | Lab model card | |
| MMSearch-Plus | 30% | #2 of 2 | Lab model card | |
| ImageMining | 30.7% | #1 of 1 | Lab model card | |
| MMSearch | 72.9% | #1 of 1 | Lab model card | |
| ZClawBench | 57.6% | #1 of 1 | Lab model card |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| RadLE 2.0 | 409.0 | #7 of 12 | Official leaderboard | |
| DesignArena Python-PPTX Slides | 1,109 | #27 of 36 | Official leaderboard | |
| DesignArena HTML Slides | 1,016 | #25 of 28 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| VisFactor | 27.2% | #10 of 26 | Official leaderboard | |
| Vision Arena | 1,232 | #60 of 136 | Official leaderboard | |
| V* | 89% | #11 of 17 | Lab model card | |
| CiteVQA | 12.8% | #13 of 16 | Official leaderboard | |
| Claw-Eval-MM | 47% | #9 of 11 | Official leaderboard | |
| BVB | 59.0 | #38 of 45 | Official leaderboard | |
| PerceptionBench (Moonshot) | 39.6% | #14 of 16 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| HarnessAudit-Bench | 0.31 | #4 of 7 | Official leaderboard |
The Self-Reported Index fits one score to the 6 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 147.1 to 157.0. 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.
The benchmarks the lab chose to headline, with the scores it gave.