No — not on this device
DeepSeek V4 Pro 1.6T-A49B at Q4_K_M needs 928.9 GB against 204.8 GB usable, and the shortfall of 724.1 GB is more than 32 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 | — | ~1.2 | −0.1% ppl | 1429.1 GB over |
| Q6_K | 1260.1 GB | 1261.2 GB | — | ~1.5 | −0.4% ppl | 1056.4 GB over |
| Q5_K_M | 1089.1 GB | 1090.3 GB | — | ~1.8 | −0.8% ppl | 885.5 GB over |
| Q4_K_M | 927.8 GB | 928.9 GB | — | ~2.1 | −1.9% ppl | 724.1 GB over |
| Q3_K_M | 751.1 GB | 752.2 GB | — | ~2.6 | −5.4% ppl | 547.4 GB over |
| Q2_K | 643.5 GB | 644.6 GB | — | ~3.0 | −15% ppl | 439.8 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
$ llama-server \
-hf deepseek-ai/DeepSeek-V4-Pro-0813:Q4_K_M \
-c 8192 -ngl 13
The engine underneath most of the others. Every knob is exposed. More on llama.cpp.