GLM-4 models
6 GLM-4 models on Hugging Face, from 9.4B to 32.6B parameters. At the precision each one is published in, the smallest needs about 21.0 GB of VRAM (glm-4-9b-hf, cheapest live fit: RTX A5000) and the largest about 72.8 GB (GLM-Z1-32B-0414, cheapest live fit: A100). The cheapest way to run glm-4-9b-hf is $0.176/hr.
GLM-4 models
| Model | Parameters | Published as | VRAM needed | Live GPU fit | Est. $/hr |
|---|---|---|---|---|---|
| glm-4-9b-hf | 9.4B | BF16 | 21.0 GB | RTX A5000 | $0.176/hr |
| codegeex4-all-9b | 9.4B | BF16 | 21.0 GB | RTX A5000 | $0.176/hr |
| glm-4-9b | 9.4B | BF16 | 21.0 GB | RTX A5000 | $0.176/hr |
| GLM-4-9B-0414 | 9.4B | BF16 | 21.0 GB | RTX A5000 | $0.176/hr |
| GLM-4.1V-9B-Thinking | 10.3B | BF16 | 23.0 GB | RTX A5000 | $0.176/hr |
| GLM-Z1-32B-0414 | 32.6B | BF16 | 72.8 GB | A100 | $1.21/hr |
VRAM is for the precision the model is published in, with the same overhead and an 8,192-token context assumed on every page; see the methodology. The fit shown is the lowest-priced single GPU type that holds the model at that precision, or the lowest-priced multi-GPU set (up to 8) when none does.
