← CanIRunLocal · All hardware · All models
No. DeepSeek V3.1 671B (MoE) does not fit on M2 Ultra 192GB: Q4_K_M needs 406.4 GB and 144.0 GB is usable.
Hardware: M2 Ultra 192GB · Model: DeepSeek V3.1 671B (MoE)
M2 Ultra 192GB has 192 GB of unified memory and 800 GB/s of memory bandwidth. The checker leaves macOS a share of unified memory, so Ollama can use 144.0 GB. DeepSeek V3.1 671B (MoE) (671B, 37B active, DeepSeek) at Q4_K_M is a 404 GB download and needs 406.4 GB with headroom for about 8k tokens.
| Best quant | Q4_K_M |
|---|---|
| Download size | 404 GB |
| Memory needed | 406.4 GB |
| Usable fast memory | 144.0 GB |
| Fit | does not fit |
| Estimated speed | no estimate, it does not fit |
| Command | ollama run deepseek-v3.1:671b |
Smallest hardware that runs it: Mac Studio M3 Ultra (80c GPU) 512GB at Q4_K_M (fits if the Mac GPU limit is raised, 406.4 GB needed, 384.0 GB usable, about 15 tok/s (estimate)). Can a Mac Studio M3 Ultra (80c GPU) 512GB run DeepSeek V3.1 671B (MoE)?
A smaller DeepSeek model that fits this hardware: DeepSeek R1 Distill Llama 70B at Q8_0 (fits in fast memory, 79.1 GB). Can a M2 Ultra 192GB run DeepSeek R1 Distill Llama 70B?
Open M2 Ultra 192GB 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