Z.AI
availableShows if the model has enough results for an index.GLM-5.3
GLM-5.3 is a reasoning model from Z.AI in the GLM-5 family. 51 benchmarks count toward its score, in 6 categories.
IndexOverall score. 50 is the middle.69.6 ±3.8
CoverageShare of the index weight with results.85%
SpeedOutput tokens per second.152/s
Input / 1MUS dollars per 1M input tokens.$0.84 batch $0.7
Output / 1MUS dollars per 1M output tokens.$2.64 batch $2.2 US dollars per 1M output tokens in a batch.
ContextMaximum tokens in one request.1.31M
EloLMArena rating and rank.1475 (#20)
50 is the middle of the board. The range shows the doubt in the index. Batch work costs less.
10,960 votes. Elo shows what people prefer. It does not change the score.
CapabilitiesScore per category. 50 is the middle.
50 is the middleResults
51 counted| BenchmarkThe test name. | CategoryThe capability that the test measures. | ResultThe score from the publisher. | IndexThis result on the index scale. | RunThe settings of the run. | DateDate of the result. | Published byThe source of the result. |
|---|---|---|---|---|---|---|
| SWE-bench | Coding | 95.4% | 74.1 | — | 1 Sept 2026 | Vals AI |
| Artificial Analysis GPQA Diamond | Knowledge | 91.7% | 62.5 | — | — | Artificial Analysis |
| OTIS Mock AIME 2024-2025 | Math | 91.1% | 62.5 | max effort | — | Epoch AI |
| GPQA diamond | Knowledge | 90.9% | 62.2 | max effort | — | Epoch AI |
| Terminal-Bench 2.1 (provider run) | Agentic | 88.2% | 76.1 | — | — | DeepSeek-AI |
| Terminal-Bench 2.1 (provider run) | Agentic | 88.2% | 76.1 | — | — | DeepSeek-AI |
| GPQA Diamond | Knowledge | 88.1% | 59.6 | — | 1 Sept 2026 | Vals AI |
| LiveBench Mathematics | Math | 87.9% | 65.3 | — | 25 Jun 2026 | LiveBench |
| MMLU Pro | Knowledge | 86.8% | 57.3 | — | 1 Sept 2026 | Vals AI |
| LiveBench Reasoning | Reasoning | 85.8% | 75.2 | — | 25 Jun 2026 | LiveBench |
| CyberGym | Agentic | 84.5% | 74.0 | — | — | Zhun Wang et al. |
| LiveCodeBench | Coding | 80.5% | 57.5 | — | 1 Sept 2026 | Vals AI |
| LiveBench Language | Knowledge | 79.9% | 66.9 | — | 25 Jun 2026 | LiveBench |
| Artificial Analysis Long Context Reasoning | Reasoning | 79.7% | 63.4 | — | — | Artificial Analysis |
| LiveBench Coding | Coding | 79.0% | 68.7 | — | 25 Jun 2026 | LiveBench |
| VulcanBench v3 | Coding | 78.3% | 58.6 | — | — | VulcanBench contributors |
| Vibe Code Bench v1.1 | Coding | 78.1% | 74.7 | OpenHands | 21 Sept 2026 | Vals AI |
| FrontierSWE | Coding | 78.1% | — | — | — | Evan Chu et al. |
| Artificial Analysis Coding Index | Coding | 74.8% | 71.6 | — | — | Artificial Analysis |
| Toolathlon-Verified | Agentic | 73.0% | 71.4 | — | — | Moonshot AI |
| Terminal-Bench 2.1 | Agentic | 71.5% | 66.3 | — | 21 Sept 2026 | Vals AI |
| LiveBench Data Analysis | Reasoning | 70.2% | 53.5 | — | 25 Jun 2026 | LiveBench |
| LiveBench Instruction Following | Instruction | 69.3% | 69.4 | — | 25 Jun 2026 | LiveBench |
| FrontierMath-Tiers-1-3-v2-Private | Math | 68.8% | 70.2 | max effort | — | Epoch AI |
| IOI | Coding | 68.4% | 75.5 | — | 21 Sept 2026 | Vals AI |
| DeepSWE | Agentic | 66.9% | 72.4 | — | — | Datacurve AI |
| Humanity's Last Exam with tools | Agentic | 62.5% | 75.1 | — | — | DeepSeek-AI |
| Artificial Analysis AutomationBench | Agentic | 62.2% | 74.9 | — | — | Artificial Analysis |
| LiveBench Agentic Coding | Agentic | 60.9% | 74.0 | — | 25 Jun 2026 | LiveBench |
| OpenHarmony Bench v1.0 | Coding | 60.8% | 70.0 | — | — | OpenHarmony Bench authors |
| Artificial Analysis SciCode | Coding | 59.0% | 74.7 | — | — | Artificial Analysis |
| NL2Repo | Coding | 58.0% | 70.6 | — | — | MiniMax |
| GDPval-AA normalized | Agentic | 57.3% | 79.6 | — | — | Artificial Analysis |
| Artificial Analysis Agentic Index | Agentic | 53.4% | 80.7 | — | — | Artificial Analysis |
| Artificial Analysis Tau3-Banking | Agentic | 50.3% | 80.9 | — | — | Artificial Analysis |
| ProofBench v1.1 | Math | 49.0% | 69.2 | — | 21 Sept 2026 | Vals AI |
| Medical Long Context Reasoning (MLCR-AA) | Reasoning | 48.3% | 80.0 | — | — | Wisedocs and Artificial Analysis |
| AutomationBench | Agentic | 48.2% | 95.0 | — | — | Moonshot AI |
| SkillsBench | Coding | 47.5% | 63.3 | OpenHands | 11 Sept 2026 | Vals AI |
| Artificial Analysis Intelligence Index | Knowledge | 44.8% | 78.4 | — | — | Artificial Analysis |
| Code Migration | Coding | 44.2% | 73.2 | — | 21 Sept 2026 | Vals AI |
| SWE-Marathon | Coding | 42.5% | — | — | — | Abundant AI and BenchFlow |
| Artificial Analysis Humanity's Last Exam | Knowledge | 42.3% | 70.4 | — | — | Artificial Analysis |
| Terminal-Bench 4.0.0 | Agentic | 41.8% | 88.9 | max effort · Claude Code | 21 Sept 2026 | Terminal-Bench |
| SimpleQA Verified | Knowledge | 41.0% | 59.5 | max effort | — | Epoch AI |
| PostTrain Bench | Coding | 39.8% | — | — | — | Moonshot AI |
| Artificial Analysis EnterpriseOps-Gym | Agentic | 36.4% | 57.2 | — | — | Artificial Analysis |
| Artificial Analysis Omniscience Accuracy | Knowledge | 33.9% | 55.8 | — | — | Artificial Analysis |
| Mystery Game Puzzles | Reasoning | 33.0% | 69.0 | max effort | — | Epoch AI |
| FrontierSWE v2 | Coding | 30.2% | 71.2 | — | — | Proximal |
| FrontierMath-Tier-4-v2-Private | Math | 29.3% | 62.0 | max effort | — | Epoch AI |
| Agents' Last Exam | Agentic | 28.5% | 68.0 | — | — | DeepSeek-AI |
| terminalBench3 | Agentic | 28.3% | — | — | — | Benchmark authors |
| terminalBench3 | Agentic | 28.3% | — | — | — | Benchmark authors |
| Terminal-Bench 4.0 | Agentic | 25.3% | 77.2 | — | 21 Sept 2026 | Vals AI |
| Chess Puzzles | Reasoning | 21.0% | 50.7 | max effort | — | Epoch AI |
| Vibe Code Bench 1-100 | Coding | 20.0% | 73.8 | OpenHands | 16 Sept 2026 | Vals AI |
| Critical Physics Tasks | Reasoning | 19.1% | 78.4 | — | — | Artificial Analysis |
| ProgramBench: Can Language Models Rebuild Programs From Scratch? | Coding | 19.0% | 34.5 | — | — | John Yang et al. |
| ExploitGym | Agentic | 15.0% | 72.0 | — | — | Zhun Wang et al. |
| Agent Arena task outcome | Agentic | 8.5 | 76.9 | max effort | 15 Sept 2026 | LMArena |
| Agent Arena steerability | Agentic | 1.9 | 69.5 | max effort | 15 Sept 2026 | LMArena |
| ProgramBench | Coding | 1.5% | — | — | 21 Sept 2026 | Vals AI |
| Agent Arena command recovery | Agentic | -1.0 | 66.2 | max effort | 15 Sept 2026 | LMArena |
51 benchmarks count, from 58 of 64 results. A grey row does not count. Too few models took that benchmark.