No — not on this device
DeepSeek-R1 671B at Q4_K_M needs 378.4 GB against 18.0 GB usable, and the shortfall of 360.4 GB is more than 16 GB of system RAM can cover at a tolerable speed. A smaller sibling or a lower quantisation is the honest answer here.
The January 2025 moment. Multi-head latent attention keeps its KV cache tiny; the weights do not.
What hardware do I need for DeepSeek-R1 671B? →
Fits instead: DeepSeek-R1-Distill-Qwen 14B (10.4 GB) · DeepSeek-R1-Distill-Qwen 7B (5.3 GB)
The VRAM budget
Quantisation ladder
| Quant | Weights | Total @ 8K | Max context | Tok/s | Quality | Fit |
|---|---|---|---|---|---|---|
| Q8_0 | 664.0 GB | 665.1 GB | — | ~0.7 | −0.1% ppl | 647.1 GB over |
| Q6_K | 512.4 GB | 513.6 GB | — | ~0.9 | −0.4% ppl | 495.6 GB over |
| Q5_K_M | 442.9 GB | 444.0 GB | — | ~1.1 | −0.8% ppl | 426.0 GB over |
| Q4_K_M | 377.3 GB | 378.4 GB | — | ~1.3 | −1.9% ppl | 360.4 GB over |
| Q3_K_M | 305.4 GB | 306.6 GB | — | ~1.5 | −5.4% ppl | 288.6 GB over |
| Q2_K | 261.7 GB | 262.8 GB | — | ~1.8 | −15% ppl | 244.8 GB over |
Quality is the published perplexity delta against f16 weights. Max context assumes an f16 KV cache; q8_0 roughly doubles it. This model uses multi-head latent attention, so its cache is a compressed latent rather than full K and V.
How to run it
$ pip install mlx-lm $ mlx_lm.generate --model mlx-community/DeepSeek-R1-4bit \ --max-tokens 512 --prompt "Hello"
Apple's own array framework. The fastest path on Apple Silicon. More on MLX.