GLM-4-32B-0414 and GLM-Z1
Open 32B base, chat and reasoning (Z1, Z1-Rumination) models, with a 9B reasoning variant, positioned against much larger GPT-4o and DeepSeek-V3.
Six open checkpoints at 9B and 32B. The 32B chat model scores 87.6 on IFEval; GLM-Z1 adds long chain-of-thought reasoning and Z1-Rumination does deeper multi-step research. Zhipu claims the 32B rivals larger models; independent checks were thin.
- Date
- Monday, 14 April 2025
- Lab
- Zhipu AI / Z.ai
- Kind
- open-weights
- Access
- open weights
Figures
| Measure | Value | Measured by |
|---|---|---|
| IFEval (GLM-4-32B-0414) | 87.6 company-reported | company |
Part of the post-DeepSeek-R1 wave of open reasoning models from Chinese labs; MIT licence per the HF model card.
Sources
This record was checked against its sources on 6 October 2026. How we check