- Terminal Bench 2.1 (Claude Code 2.1.207) 84.3
- DeepSWE (v1.1) 63.4
- NL2Repo 56.3
- Toolathlon Verified 78.4
- AutomationBench (v1.0.6) 48.8
- Agents Last Exam 26.3
- HLE w/ Tools (full set) 55.3
- GDPval-AA v2 1773
- OfficeQA Pro (Treasury Bulletin PDFs, no embedded text) 62.4
- CharXiv Reasoning (w/ Tools) 89.4
- Chartography (w/ Tools) 78
- BabyVision 53.4
- MVbench 77.8
- MMVU 80.5
- OSWorld 2.0 59.1
- Vision2Web 77.8
- Z.ai Code Bench v1.0, 29.0 29
- AA Intelligence Index v4.1.1, 57 57
GLM-5.3-Flash
GLM-5.3-Flash ranks first on 3 of the 165 benchmarks with five or more models scored.
165
Benchmarks with a score
3
Ranked first, out of five or more models
9
In the top three
Z.ai Aug 2026
GLM-5.3-Flash
Open weights
153.8
#52 of 256
Epoch Capabilities Index
Self-Reported Index 158.8 #39 of 404
- DeepSWE 63.4% #13 /28
- NL2Repo-Bench 56.3% #10 /30
- Terminal-Bench 4.0 36.7% #18 /40
Where it ranks
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.
Knowledge and language
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| LMArena (Elo) | 1,474 | #33 of 335 | Official leaderboard | |
| Humanity's Last Exam | 55.3% | #25 of 154 | Lab model card | |
| HABIT-Bench | 42% | #1 of 6 | Official leaderboard | |
| DesignArena ASCII Art | 1,269 | #13 of 77 | Official leaderboard | |
| OpenAI MRCR v2 | 60.1% | #11 of 47 | Official leaderboard | |
| AgentIF (2026-10 setting) | 64.4% | #8 of 24 | Official leaderboard | |
| AA-LCR Extended | 66.9% | #20 of 49 | Official leaderboard | |
| Hemingway-bench | 1,034 | #20 of 42 | Official leaderboard | |
| StudyArena Text | 1,227 | #49 of 85 | Official leaderboard | |
| FACTS Grounding v2 | 56.1% | #22 of 25 | Official leaderboard |
Math and science
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| GPQA Diamond | 90.2% | #47 of 229 | Epoch AI run | |
| OTIS Mock AIME 2024-2025 | 93.9% | #56 of 212 | Epoch AI run | |
| Verifying the Verifiers (Project APE CRED benchmark) | 90% | #7 of 23 | Official leaderboard | |
| ARC-AGI-2 | 65.8% | #30 of 98 | Official leaderboard | |
| PHYSICS-Verified | 85.3% | #6 of 19 | Official leaderboard | |
| ARC-AGI | 91% | #31 of 95 | Official leaderboard | |
| NatureBench | 11.1% | #10 of 18 | Official leaderboard | |
| FrontierMath Tiers 1-3 v2 | 55.8% | #48 of 86 | Official leaderboard | |
| Surface Evolver Bench | 52.5% | #15 of 25 | Official leaderboard | |
| FrontierMath Tier 4 v2 | 17.1% | #51 of 68 | Official leaderboard | |
| ProofBench v1.1 | 21% | #41 of 52 | Official leaderboard | |
| PhysClosureBench | 3.28 | #6 of 7 | Official leaderboard | |
| BTF-3 | 0.15 | #7 of 8 | Official leaderboard | |
| Science's Last Exam | 21.3 | #8 of 9 | Official leaderboard |
Coding
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| DesignArena UI Component | 1,298 | #24 of 139 | Official leaderboard | |
| WebDev Arena | 1,607 | #19 of 105 | arena.ai, via Epoch AI | |
| DesignArena SVG | 1,272 | #20 of 103 | Official leaderboard | |
| DesignArena 3D Design | 1,296 | #27 of 136 | Official leaderboard | |
| Code Arena | 1,609 | #30 of 128 | Official leaderboard | |
| DesignArena Code | 1,268 | #36 of 143 | Official leaderboard | |
| DesignArena Game Dev | 1,267 | #36 of 143 | Official leaderboard | |
| SWE-bench Verified Extended | 66% | #12 of 44 | Official leaderboard | |
| DesignArena Website | 1,258 | #49 of 151 | Official leaderboard | |
| DesignArena Data Visualization | 1,257 | #39 of 118 | Official leaderboard | |
| NL2Repo-Bench | 56.3% | #10 of 30 | Lab model card | |
| StructEval | 76.2% | #10 of 28 | Official leaderboard | |
| Image-to-WebDev Arena | 1,582 | #21 of 55 | Official leaderboard | |
| APEX-SWE | 46.1% | #25 of 65 | Official leaderboard | |
| Vision2Web | 77.8% | #4 of 10 | Official leaderboard | |
| FDE-Gym | 0.22 | #5 of 12 | Official leaderboard | |
| DesignArena Video to Website | 1,195 | #8 of 19 | Official leaderboard | |
| DeepSWE | 63.4% | #13 of 28 | Official leaderboard | |
| PostTrainBench 1.2 | 36.8% | #10 of 21 | Official leaderboard | |
| ReactBench | 32.8% | #14 of 28 | Official leaderboard | |
| DesignArena Image to HTML | 1,207 | #26 of 51 | Official leaderboard | |
| IOI v2 (Vals) | 52.5% | #24 of 45 | Official leaderboard | |
| Code Migration v1.1 | 51.6% | #16 of 30 | Official leaderboard | |
| ALE-Bench | 1,283 | #74 of 126 | Official leaderboard | |
| DAXBench v2 | 75.8% | #28 of 47 | Official leaderboard | |
| Vibe Code Bench | 30.8% | #79 of 120 | Official leaderboard | |
| Code Migration | 20.5% | #52 of 78 | Official leaderboard | |
| Chart Arena | 1,469 | #10 of 15 | Official leaderboard | |
| FrontierCS 2 | 12.5% | #13 of 19 | Official leaderboard | |
| Vibe Code Bench 1-100 | 16% | #18 of 26 | Official leaderboard | |
| SWE Atlas v1.1 | 55.3% | #21 of 28 | Official leaderboard | |
| FrontierSWE | 18.1% | #20 of 25 | Official leaderboard | |
| SWE-Marathon v1.1 | 38.8% | #16 of 20 | Official leaderboard | |
| Z.ai Code Bench | 29% | #4 of 5 | Official leaderboard | |
| CursorBench | 36.8% | #18 of 21 | cursor.com, via Epoch AI | |
| CursorBench 4.0 | 36.8% | #18 of 21 | Official leaderboard | |
| FrontierCode 2.0 | 14.9% | #28 of 32 | Official leaderboard | |
| Brokk Power Ranking | 1.70 | #18 of 19 | Official leaderboard | |
| SWE-Clar | 39% | #9 of 9 | Official leaderboard |
Agents and tools
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Toolathlon-Verified | 78.4% | #1 of 25 | Official leaderboard | |
| AppWorld Test-Challenge | 92.8% | #1 of 12 | Official leaderboard | |
| AppWorld Test-Normal | 93.5% | #2 of 12 | Official leaderboard | |
| LLM Chess | 572 | #39 of 123 | Official leaderboard | |
| AutomationBench | 48.8% | #15 of 47 | Lab model card | |
| Terminal-Bench 2.1 | 84.3% | #19 of 56 | Lab model card | |
| Terminal-Bench 4.1 | 37% | #19 of 43 | Official leaderboard | |
| Terminal-Bench 4.0 | 36.7% | #18 of 40 | Official leaderboard | |
| OSWorld 2.0 | 59.1% | #10 of 19 | Lab model card | |
| GAIA | 43.9% | #35 of 64 | Official leaderboard | |
| DuelLab GameBench 2 | 60.8 | #36 of 57 | Official leaderboard | |
| Terminal-Bench 2.1 Extended | 30% | #28 of 44 | Official leaderboard | |
| Agent Arena | -2.95% | #38 of 56 | Official leaderboard | |
| BabaIsBench | 67% | #23 of 32 | Official leaderboard | |
| Chess Puzzles | 21% | #69 of 93 | Epoch AI run | |
| RuneBench | 3.83 | #57 of 76 | Official leaderboard | |
| CivBench | 1,468 | #15 of 19 | Official leaderboard | |
| Hermes Bench | 68.9% | #14 of 17 | Official leaderboard | |
| Mystery Game Puzzles | 8.00% | #75 of 84 | Epoch AI run | |
| Web Context (Nimble) | 64.2% | #11 of 11 | Official leaderboard |
Professional work
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| MedScribe | 88.9% | #9 of 97 | Official leaderboard | |
| GDPval-AA v2 | 1,773 | #3 of 27 | Lab model card | |
| TaxEval v2 | 75.6% | #16 of 129 | Official leaderboard | |
| Vals Finance Agent v2 | 57.9% | #14 of 80 | Official leaderboard | |
| Legal Research Bench | 45.2% | #23 of 93 | Official leaderboard | |
| LegalBench | 83.9% | #45 of 138 | Official leaderboard | |
| Tax Agent Bench | 31.6% | #29 of 84 | Official leaderboard | |
| CorpBench Work | 84.6 | #6 of 17 | Official leaderboard | |
| Harvey LAB (Legal Agent Benchmark) | 6.67% | #32 of 80 | Official leaderboard | |
| Vals Legal Index | 71% | #16 of 38 | Official leaderboard | |
| BlenderWorld | 59.7 | #6 of 13 | Official leaderboard | |
| APEX-Agents | 52.8% | #30 of 61 | Collected by Epoch AI, source not named | |
| FrontierCAD Parts | 0.33 | #3 of 6 | Official leaderboard | |
| APEX-Agents 1.1 | 66.8% | #33 of 65 | Official leaderboard | |
| EnterpriseBench: CoreCraft | 46.7% | #26 of 51 | Official leaderboard | |
| DAYJOB: Finance | 7.30% | #20 of 39 | Official leaderboard | |
| E-Commerce Bench | 2.58 | #11 of 19 | Official leaderboard | |
| Vals Index v2 | 47.2% | #38 of 65 | Official leaderboard | |
| MedXpertQA MM Extended | 52.9% | #24 of 41 | Official leaderboard | |
| Harvey LAB v1.1 (HLAB-Vals-1.1) | 10.4% | #23 of 38 | Official leaderboard | |
| Vals Finance Agent v2.1 | 60.6% | #25 of 40 | Official leaderboard | |
| Excel Modeling Benchmark | 55.9% | #48 of 76 | Official leaderboard | |
| Vals Finance Index | 55.8% | #27 of 40 | Official leaderboard | |
| Clipping Bench | $399 | #13 of 19 | Official leaderboard | |
| OfficeQA Pro | 62.4% | #11 of 16 | Lab model card | |
| Tuesday Work Index | 42.6 | #32 of 44 | Official leaderboard | |
| Thrive Holdings helpdesk ticket resolution benchmark | 42.3% | #11 of 15 | Official leaderboard | |
| Excel Modeling Benchmark v1.1 | 66.3% | #31 of 42 | Official leaderboard | |
| DAYJOB: Healthcare | 1.20% | #28 of 37 | Official leaderboard | |
| Agents' Last Exam | 26.3% | #20 of 26 | Lab model card | |
| Holdings-IndividualTaxBench | 71.3% | #10 of 13 | Official leaderboard | |
| APEX-Accounting | 53.7% | #38 of 49 | Official leaderboard | |
| APEX-1 | 58.7% | #56 of 71 | Official leaderboard | |
| FoodTruck Bench | $1,245 | #47 of 59 | Official leaderboard | |
| Vals Index v2.1 | 44.6% | #41 of 49 | Official leaderboard | |
| Vals Index v2.2 | 50.9% | #31 of 37 | Official leaderboard | |
| Legora BAR | 94.0 | #15 of 17 | Official leaderboard | |
| OR-License | 62.6% | #10 of 11 | Official leaderboard | |
| Hebbia Financial Services Benchmark | 43.7% | #10 of 10 | Official leaderboard | |
| FrontierCAD Assemblies | 0.08 | #6 of 6 | Official leaderboard | |
| SimuVerity | 4.98 | #5 of 5 | Official leaderboard |
Vision, audio and robotics
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| MVBench | 77.8% | #2 of 23 | Lab model card | |
| MMVU | 80.5% | #3 of 26 | Lab model card | |
| CharXiv | 89.4% | #10 of 75 | Lab model card | |
| ParseBench | 70.5% | #8 of 44 | Official leaderboard | |
| Vision Arena | 1,272 | #29 of 136 | Official leaderboard | |
| MineBench | 1,763 | #22 of 75 | Official leaderboard | |
| BVB | 67.9 | #19 of 45 | Official leaderboard | |
| VoxelBench | 1,742 | #27 of 56 | Official leaderboard | |
| CineRecon | 47.6 | #6 of 12 | Official leaderboard | |
| InteractiveBench | 30% | #6 of 12 | Official leaderboard | |
| BabyVision | 53.4% | #16 of 26 | Lab model card | |
| ExtractBench | 80.8% | #18 of 27 | Official leaderboard | |
| 4DCodeBench | 0.47 | #11 of 16 | Official leaderboard | |
| Chartography | 16.3% | #30 of 43 | Official leaderboard | |
| Humanity's Sixth Sense | 26.8% | #24 of 34 | Official leaderboard | |
| Code2Video-Compass | 10.5 | #11 of 15 | Official leaderboard | |
| ScienceVision | 0.28 | #9 of 12 | Official leaderboard | |
| DataCenterBench Tier 1 | 15% | #7 of 9 | Official leaderboard | |
| Code2Video Benchmark | 1,467 | #15 of 19 | Official leaderboard | |
| CharXiv Extended | 76.2% | #31 of 38 | Official leaderboard | |
| Frontier3DBench | 47.6 | #9 of 11 | Official leaderboard | |
| MMMU-Pro Extended | 22.6% | #35 of 38 | Official leaderboard | |
| Code2Video Arena | 1,377 | #13 of 13 | Official leaderboard | |
| Embodied First Exam | 2.41% | #9 of 9 | Official leaderboard | |
| VibeVideoBench | 2.79 | #9 of 9 | Official leaderboard |
Safety and security
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| nocot-bench (NCRI) | 76.4 | #71 of 185 | Official leaderboard | |
| DistressBench | 72.7% | #18 of 42 | Official leaderboard | |
| CyberBench (Vals) | 71.5% | #12 of 26 | Official leaderboard | |
| SecIT Bench | 75.5% | #12 of 26 | Official leaderboard | |
| CyberBench v1.1 (Vals) | 67.9% | #24 of 44 | Official leaderboard | |
| FORTRESS | 26.7% | #42 of 73 | Official leaderboard | |
| DeepsecBench | 12.9 | #22 of 29 | Official leaderboard | |
| SRE-Bench | 0.76% | #28 of 35 | Official leaderboard | |
| Biopharma Bench | 42.5% | #12 of 13 | Official leaderboard | |
| CyberOffense | 20% | #6 of 6 | Official leaderboard |
Other
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| SAGE v1.1 (Vals) | 89.3% | #5 of 29 | Official leaderboard | |
| Xbench | 52.9 | #3 of 17 | Official leaderboard | |
| SuperCLUE | 47.2 | #7 of 16 | Official leaderboard | |
| Artificial Analysis Intelligence Index v4.1 | 57.0 | #4 of 8 | Lab model card | |
| LiveBench | 60.7% | #22 of 31 | Official leaderboard | |
| Hermes Index | 35.0 | #13 of 17 | Official leaderboard |
Its lab's numbers put it at 158.8, 5.0 points above its ECI
The Self-Reported Index fits one score to the 14 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 155.6 to 162.0. How we compute it.
- CharXiv 89.4
- Terminal-Bench 2.1 84.3
- MMVU 80.5
- Toolathlon-Verified 78.4
- Chartography 78
- MVBench 77.8
- DeepSWE 63.4
- OfficeQA Pro 62.4
- OSWorld 2.0 59.1
- NL2Repo-Bench 56.3
- Humanity's Last Exam 55.3
- BabyVision 53.4
- AutomationBench 48.8
- Agents' Last Exam 26.3
Its lab's numbers next to 5 leaderboards
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 |
|---|---|---|---|
| Chartography | 78% Chartography (w/ Tools) | 16.3% Leaderboard, max | +62 pts |
| DeepSWE | 63.4% DeepSWE (v1.1) | 63.4% Leaderboard, max, mini-swe-agent | Same |
| Toolathlon-Verified | 78.4% Toolathlon Verified | 78.4% Leaderboard, max | Same |
| Vision2Web | 77.8% Vision2Web | 77.8% Leaderboard, Claude Code | Same |
| Z.ai Code Bench | 29% Z.ai Code Bench v1.0, 29.0 | 29% Leaderboard, max, Claude Code 2.1.207 | Same |
Z.ai's models over time
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)
Its launch post
The benchmarks the lab chose to headline, with the scores it gave.