No — not on this device
DeepSeek V4 Pro 1.6T-A49B at Q4_K_M needs 928.9 GB against 384.0 GB usable, and the shortfall of 544.9 GB is more than 128 GB of system RAM can cover at a tolerable speed. A smaller sibling or a lower quantisation is the honest answer here.
The 0813 refresh of the V4 flagship. Included as the honest ceiling; a terabyte of weights at Q4.
What hardware do I need for DeepSeek V4 Pro 1.6T-A49B? →
Fits instead: DeepSeek V4 Flash 284B-A13B (160.7 GB)
The VRAM budget
Quantisation ladder
| Quant | Weights | Total @ 8K | Max context | Tok/s | Quality | Fit |
|---|---|---|---|---|---|---|
| Q8_0 | 1632.7 GB | 1633.9 GB | — | ~4.3 | −0.1% ppl | 1249.9 GB over |
| Q6_K | 1260.1 GB | 1261.2 GB | — | ~5.6 | −0.4% ppl | 877.2 GB over |
| Q5_K_M | 1089.1 GB | 1090.3 GB | — | ~6.4 | −0.8% ppl | 706.3 GB over |
| Q4_K_M | 927.8 GB | 928.9 GB | — | ~7.6 | −1.9% ppl | 544.9 GB over |
| Q3_K_M | 751.1 GB | 752.2 GB | — | ~9.4 | −5.4% ppl | 368.2 GB over |
| Q2_K | 643.5 GB | 644.6 GB | — | ~11 | −15% ppl | 260.6 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-V4-Pro-0813-4bit \ --max-tokens 512 --prompt "Hello"
Apple's own array framework. The fastest path on Apple Silicon. More on MLX.