← CanIRunLocal · All hardware · All models
No. Mixtral 8x7B (MoE) does not fit on M1 Max (24c GPU) 32GB: Q4_K_M needs 29.8 GB and 24.0 GB is usable.
Hardware: M1 Max (24c GPU) 32GB · Model: Mixtral 8x7B (MoE)
M1 Max (24c GPU) 32GB has 32 GB of unified memory and 400 GB/s of memory bandwidth. M1 Max (32c GPU) 32GB and M2 Max (30c GPU) 32GB and M2 Max (38c GPU) 32GB have the same memory and bandwidth, so the fit numbers match; this page is M1 Max (24c GPU) 32GB. The checker leaves macOS a share of unified memory, so Ollama can use 24.0 GB. Mixtral 8x7B (MoE) (46.7B, 12.9B active, Mistral) at Q4_K_M is a 28 GB download and needs 29.8 GB with headroom for about 8k tokens.
| Best quant | Q4_K_M |
|---|---|
| Download size | 28 GB |
| Memory needed | 29.8 GB |
| Usable fast memory | 24.0 GB |
| Fit | does not fit |
| Estimated speed | no estimate, it does not fit |
| Command | ollama run mixtral:8x7b-instruct-v0.1-q4_K_M |
Smallest hardware that runs it: RTX 3090 24GB at Q4_K_M (partial offload (slower), 29.8 GB needed, 23.5 GB usable, about 10 tok/s (estimate)). Can a RTX 3090 24GB run Mixtral 8x7B (MoE)?
A smaller Mistral model that fits this hardware: Devstral Small 2 24B at Q4_K_M (fits in fast memory, 16.8 GB). Can a M1 Max (24c GPU) 32GB run Devstral Small 2 24B?
Open M1 Max (24c GPU) 32GB in the checkerEstimates, 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