← CanIRunLocal · All hardware · All models
Z.ai · 320B parameters (18B active) · chat, reasoning, coding, vision.
Ollama lists glm-5.3-flash:cloud as a cloud model, so there is no local download. The fits below use verified Hugging Face weights.
Yes on 8 machines in this list. The smallest is Mac Studio M3 Ultra 256GB: Q4_K_M, about 206.3 GB, roughly 29 tokens per second, via LM Studio. lms get mlx-community/GLM-5.3-Flash-4bit
lms get mlx-community/GLM-5.3-Flash-4bitllama-server -hf unsloth/GLM-5.3-Flash-GGUF:Q8_0 --rpc node2:50052,node3:50052,node4:50052 -c 8192 # run rpc-server on each other nodeexo # on every Mac (brew install --cask exo); load mlx-community/GLM-5.3-Flash-4bit with Tensor sharding + RDMA| Hardware | Runtime | Quant | Memory | Fit | Speed | Command |
|---|---|---|---|---|---|---|
| Mac Studio M3 Ultra 256GB | LM Studio | MLX 4bit | 206.3 GB | fits if the Mac GPU limit is raised | ~29 tok/s | lms get mlx-community/GLM-5.3-Flash-4bit |
| Mac Studio M5 Ultra (64c GPU) 256GB | LM Studio | MLX 4bit | 206.3 GB | fits if the Mac GPU limit is raised | ~42 tok/s | lms get mlx-community/GLM-5.3-Flash-4bit |
| Mac Studio M5 Ultra (80c GPU) 256GB | LM Studio | MLX 4bit | 206.3 GB | fits if the Mac GPU limit is raised | ~42 tok/s | lms get mlx-community/GLM-5.3-Flash-4bit |
| 4x Mac Studio M5 Max (40c GPU) 128GB | llama.cpp | Q8_0 GGUF | 343.3 GB | fits in fast memory | ~10 tok/s | llama-server -hf unsloth/GLM-5.3-Flash-GGUF:Q8_0 --rpc node2:50052,node3:50052,node4:50052 -c 8192 # run rpc-server on each other node |
| 2x Mac Studio M5 Ultra (80c GPU) 256GB | llama.cpp | Q8_0 GGUF | 343.3 GB | fits in fast memory | ~21 tok/s | llama-server -hf unsloth/GLM-5.3-Flash-GGUF:Q8_0 --rpc node2:50052 -c 8192 # run rpc-server on each other node |
| Mac Studio M3 Ultra (80c GPU) 512GB | LM Studio | MLX 4bit | 206.3 GB | fits in fast memory | ~29 tok/s | lms get mlx-community/GLM-5.3-Flash-4bit |
| 4x DGX Spark 128GB | llama.cpp | Q8_0 GGUF | 343.3 GB | fits in fast memory | ~6 tok/s | llama-server -hf unsloth/GLM-5.3-Flash-GGUF:Q8_0 --rpc node2:50052,node3:50052,node4:50052 -c 8192 # run rpc-server on each other node |
| 4x Mac Studio M5 Ultra (80c GPU) 256GB | llama.cpp | Q8_0 GGUF | 343.3 GB | fits in fast memory | ~19 tok/s | llama-server -hf unsloth/GLM-5.3-Flash-GGUF:Q8_0 --rpc node2:50052,node3:50052,node4:50052 -c 8192 # run rpc-server on each other node |
Estimates, not benchmarks. Model sizes are ollama.com download sizes; memory needed adds headroom for ~8k tokens of context; speeds are rough figures from memory bandwidth. Data updated 2026-10-16.
A project by OnDevice Labs · llms.txt