GLM-5.3: benchmarks and ranking
GLM-5.3 (Z.ai (Zhipu)) — overall rank: #31 / 153, overall score: 76.7 (90% interval 72.2–80.7), leaderboards: 6. Released: 2026-08-14 · Price (in/out): $1.4 / $4.4
Rank on each leaderboard
| Leaderboard | Rank | Published value |
|---|---|---|
| Artificial Analysis | #17 / 141 | 44.8 |
| LiveBench | #31 / 64 | 76.1 |
| Epoch AI | #34 / 230 | 157.1 |
| ARC Prize (ARC-AGI) | Not rated | — |
| Scale AI SEAL | Not rated | — |
| Vals AI | #19 / 42 | 59.8 |
| Terminal-Bench | #20 / 53 | 41.21 |
| EQ-Bench | Not rated | — |
| SimpleBench | #34 / 112 | 66.2 |
| BenchLM.ai | Not rated | — |
Domain scores
- Coding: 80.7
- Agents: 74.8
- Reasoning: 73.5
- Math: 64.5
- Knowledge: 78.6
- Writing: 73.4
- Instructions: 65.6
- Chat & EQ: 59.4
- Business: 77.5
- Japanese: 82.3
- Long context: 50.3