Model family · 7 sizes
GLM: which size runs locally?
GLM comes in 7 sizes, from 9B to 744B. Some sizes are Mixture-of-Experts, so they run faster than their memory footprint suggests. Here is each size with its Q4_K_M weight, the memory it needs, and the hardware that runs it.
- Sizes
- 7
- Smallest
- 9B
- Largest
- 744B
- Runs from
- 16GB
The GLM lineup
"Needs" is the sourced minimum memory for Q4_K_M with a small context. Larger context needs more.
Which GLM fits your memory
No GLM size fits 8GB; even GLM-4 9B needs more.
Largest that fits: GLM-4-9B-0414 (9B), best case on Nvidia GeForce RTX 4080 (16GB).
Largest that fits: GLM-4-32B-0414 (32B), best case on Nvidia GeForce RTX 4090 (24GB).
Largest that fits: GLM-4-32B-0414 (32B), best case on Nvidia GeForce RTX 5090 (32GB).
Largest that fits: GLM-4-32B-0414 (32B), best case on Apple M5 Pro (48GB).
Largest that fits: GLM-4-32B-0414 (32B), best case on Apple M4 Max (64GB).
Largest that fits: GLM-4.5-Air (106B), best case on Apple M5 Max (128GB).
Largest that fits: GLM-4.5-Air (106B), best case on Apple M3 Ultra (256GB).
Best case means the most capable device at that size (usually a discrete GPU). A Mac at the same size sits roughly one rung lower; see the per-size breakdown on each memory budget page.
FAQ
Which GLM size should I run locally?
Pick the largest size your memory allows. On 16GB (best case) up to GLM-4-9B-0414; On 24GB (best case) up to GLM-4-32B-0414; On 32GB (best case) up to GLM-4-32B-0414; On 48GB (best case) up to GLM-4-32B-0414; On 64GB (best case) up to GLM-4-32B-0414; On 128GB (best case) up to GLM-4.5-Air; On 256GB (best case) up to GLM-4.5-Air. Smaller sizes run faster and leave headroom for context.
What is the smallest GLM model?
GLM-4 9B at 9B parameters, about 5.82 GB on disk at Q4_K_M and roughly 8 GB of memory to run. It is the one to use on phones and 8 GB machines.
What is the largest GLM model and what does it need?
GLM-5.2 at 744B (mixture of experts), about 465.83 GB at Q4_K_M. It needs more than a typical 32 GB desktop; a high-memory Mac or multi-GPU rig.
Understand the numbers
Short guides to the ideas behind GLM's memory and quant figures.
Sources
- docs.z.ai
- github.com
- huggingface.co/bartowski/glm-4-9b-chat-GGUF
- huggingface.co/bartowski/THUDM_GLM-4-32B-0414-GGUF
- huggingface.co/bartowski/THUDM_GLM-4-9B-0414-GGUF
- huggingface.co/THUDM/GLM-4-32B-0414
- huggingface.co/THUDM/GLM-4-9B-0414
- huggingface.co/THUDM/glm-4-9b-chat
- huggingface.co/unsloth
- huggingface.co/zai-org/GLM-4-32B-0414
- huggingface.co/zai-org/GLM-4-9B-0414
- huggingface.co/zai-org/GLM-4.5-Air
- huggingface.co/zai-org/GLM-4.7-Flash
- ollama.com/library/glm-4.7-flash
- ollama.com/library/glm4
- ollama.com/library/glm4/tags
Memory figures are estimates at Q4_K_M. See methodology.