Z.ai
8 ranked models in a catalog of 20 text models. GLM-5.3 leads this lab at #24 overall.
Ranked models
Global ranks · current snapshot| Rank | Model | SI Score | coding | math | reasoning | preference | Confidence | Blended price | Context | Released |
|---|---|---|---|---|---|---|---|---|---|---|
| #24 | GLM-5.3Open weights | 67.7 | 63.2 | 74.8 | 84.4 | 79.5 | 93% | $2.15 | 1M | Aug 14, 2026 |
| #30 | GLM-5.3-FlashOpen weights | 66.2 | 62.0 | 65.6 | 73.3 | 79.5 | 100% | $0.24 | 1M | Aug 26, 2026 |
| #34 | GLM-5.2Open weights | 65.7 | 66.2 | 70.4 | 68.3 | 79.4 | 100% | $2.15 | 1M | Jun 13, 2026 |
| #43 | GLM-5.1Open weights | 63.3 | 74.2 | 51.7 | 89.9 | 78.7 | 64% | $2.15 | 200K | Apr 7, 2026 |
| #57 | GLM-4.7Open weights | 61.3 | 53.6 | 83.3 | 83.3 | 76.4 | 69% | $1.00 | 205K | Dec 22, 2025 |
| #65 | GLM-5Open weights | 59.4 | 72.4 | 80.0 | 35.0 | 77.4 | 90% | $1.55 | 205K | Feb 12, 2026 |
| #97 | GLM-4.7-FlashOpen weights | 53.3 | 59.2 | 41.7 | 52.8 | 68.0 | 69% | $0.00 | 200K | Jan 19, 2026 |
| #114 | GLM-4.6Open weights | 47.8 | 33.0 | — | — | 76.8 | 55% | $1.00 | 205K | Sep 30, 2025 |
Pillars use a 0–100 scale. * marks an imputed neutral prior where the model has no scored results in that pillar; it is not a measured benchmark result. Prices are USD per 1M tokens; see the blend and price comparison.
Scores and releases
Provisional models
12 awaiting rank-eligible evidence- GLM-5V-Turbo
Results cover 1 pillar; at least 2 required
- GLM-5-Turbo
No usable open benchmark results found
- GLM-4.7-FlashX
No usable open benchmark results found
- GLM-4.6V
Results cover 1 pillar; at least 2 required
- GLM-4.6V-Flash
No usable open benchmark results found
- GLM-4.5V
Evidence completeness 37.0%; at least 50% required
- GLM-4.5
Evidence completeness 47.0%; at least 50% required
- GLM-4.5-Air
Evidence completeness 37.0%; at least 50% required
- GLM-4.5-Flash
No usable open benchmark results found
- GLM-4.5-AirX
No usable open benchmark results found
- GLM-4.5-X
No usable open benchmark results found
- GLM-5.3-FlashX
No usable open benchmark results found