GLM-5.3 Flash (Max): benchmark results
Provider: Zhipu. Released 2026-08-20. Access: Open.
Aggregate Rating 1781 ± 10, rank #172 of 2168 rated models, from 173 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Nejumi 4 - MT-Bench (Japanese) - Extraction | 99 | Judge rating (1-10, x10) | 98.3 |
| Nejumi 4 - jaster (2-shot) - JSQuAD | 82.06 | Character F1 (x100) | 97.9 |
| Nejumi 4 - jaster (0-shot) - JBLiMP | 97 | Exact match (%) | 95.5 |
| Nejumi 4 - jaster (2-shot) - JCommonsenseQA | 100 | Exact match (%) | 94.8 |
| Vals AI MedScribe | 88.94 | Accuracy (%) | 92.7 |
| Nejumi 4 - MT-Bench (Japanese) - Roleplay | 99 | Judge rating (1-10, x10) | 92 |
| Nejumi 4 - MT-Bench (Japanese) - Coding | 100 | Judge rating (1-10, x10) | 91.3 |
| APEX-Harm - Task Utility - Static Script Injection | 0.85 | Mean Task Score (0–1) | 90.9 |
| Nejumi 4 - MT-Bench (Japanese) - Humanities | 100 | Judge rating (1-10, x10) | 90.9 |
| Nejumi 4 - JBBQ (2-shot) - Accuracy | 94 | Accuracy (%) | 90.2 |
| Nejumi 4 - MT-Bench (Japanese) - STEM | 100 | Judge rating (1-10, x10) | 90.2 |
| Vals AI SWE-bench Verified | 92 | Resolved (%) | 89.7 |
Interactive version: theaggregate.ai/model?slug=glm-5-3-flash-max · How It Works · Data refreshed daily, snapshot 2026-10-20.