← CanIRunLocal · All hardware · All models
Yes. Qwen 2.5 Coder 32B fits in fast memory on M3 Max (40c GPU) 48GB at Q5_K_M (25.2 GB needed, 36.0 GB usable), about 14 tok/s (estimate).
Hardware: M3 Max (40c GPU) 48GB · Model: Qwen 2.5 Coder 32B
M3 Max (40c GPU) 48GB has 48 GB of unified memory and 400 GB/s of memory bandwidth. The checker leaves macOS a share of unified memory, so Ollama can use 36.0 GB. Qwen 2.5 Coder 32B (32.8B, Alibaba) at Q5_K_M is a 23 GB download and needs 25.2 GB with headroom for about 8k tokens.
| Best quant | Q5_K_M |
|---|---|
| Download size | 23 GB |
| Memory needed | 25.2 GB |
| Usable fast memory | 36.0 GB |
| Fit | fits in fast memory |
| Estimated speed | ~14 tok/s (estimate) |
| Command | ollama run qwen2.5-coder:32b-instruct-q5_K_M |
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 2026-10-23.
A project by OnDevice Labs · llms.txt