Loading…
Loading…
Models / GLM-5.3
1 model (~753B) · ↓ 1.9M downloads a month · from $11.77/h on 8x RTX PRO 6000 right now. Each model's versions do the same job; quantized ones (AWQ, FP8, 4-bit) need less GPU memory and run on cheaper machines.
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| Originalzai-org/GLM-5.3 | vLLM | 756 GB | 876.7 GB | $44.03/h 8x Nvidia H200 |
| NVFP4 (nvidia)nvidia/GLM-5.3-NVFP4 | vLLM | 464 GB | 529.6 GB | $11.77/h 8x RTX PRO 6000 |
| colibri int4 g64 (Justvugg)Justvugg/GLM-5.3-colibri-int4-g64 | vLLM | 419 GB | 503 GB | $11.77/h 8x RTX PRO 6000 |
| Quark MXFP4 AttnFP8 (amd)amd/GLM-5.3-Quark-MXFP4-AttnFP8 | vLLM | 410 GB | 492.1 GB | $11.77/h 8x RTX PRO 6000 |
| NVFP4 (incoai)incoai/GLM-5.3-NVFP4 | vLLM | 465 GB | 553.6 GB | $11.77/h 8x RTX PRO 6000 |
Model details come from Hugging Face and are refreshed daily; prices follow the live offers. GPU memory is GridPup's estimate: the weights, room for an 8,192-token conversation (chat models) and the runtime's own memory. Large models are split across the GPUs of one machine ("4x RTX 4090").