How to Deploy DeepSeek V4 Pro Locally
DeepSeek V4 Pro is a MoE model with 1,600B total parameters and 49B active per token. At Q4 it needs 894GB of VRAM to sit entirely on the GPU; with experts offloaded to system RAM it needs only 17GB of VRAM plus 879GB of RAM. No single card in the database fits it; it needs multiple GPUs or the unified memory of a Mac Studio. In offload mode the cheapest card that runs it is the GeForce RTX 2080 Ti 22GB, at roughly 3 tokens/s.
DeepSeek V4 Pro details
How much VRAM does DeepSeek V4 Pro need for local deployment?
MoE: all 1,600B parameters must stay resident in memory, while speed is set by the 49B active per token.
| Quantization | WeightsGB | Total at 8KGB | Total at 32KGB | Total at 128KGB |
|---|---|---|---|---|
| 1-bit | 375.5est. | 377.5 | 380.3 | 391.8 |
| 2-bit | 594.6est. | 596.5 | 599.4 | 610.8 |
| 3-bit | 766.7est. | 768.6 | 771.5 | 782.9 |
| 4-bit | 849.7GGUF | 893.8 | 896.6 | 908.1 |
| 5-bit | 1,079.6est. | 1,081.5 | 1,084.4 | 1,095.8 |
| 8-bit | 873.4GGUF | 1,660.5 | 1,663.3 | 1,674.7 |
| FP16 | 3,129.2est. | 3,131.2 | 3,134.1 | 3,145.5 |
Total = weights + KV cache (FP16) + 1GB runtime overhead. Weights marked GGUF are measured file sizes from the repository; the rest are estimated from bytes per parameter. Contexts beyond this model's 1M limit show a dash. Each tier takes whichever file of that bit width the repository has, preferring unsloth dynamic quants, so the exact file name varies by model.
Which GPUs can run DeepSeek V4 Pro?
Xianyu
No single card in the database fits DeepSeek V4 Pro: it needs at least 894GB of VRAM at Q4_K_M with an 8K context.
A multi-GPU build needs at least 894GB of VRAM in total.
None of the unified-memory machines in the database fit it either.
Recommended GPUs for DeepSeek V4 Pro with CPU/GPU offloading
Xianyu
Expert weights live in system RAM (needs at least 879 GB); VRAM holds only attention layers, shared experts and the KV cache, at least 17 GB at Q4. Speed is bound by RAM bandwidth (estimated at 70 GB/s) and is far slower than a full-VRAM setup.
Deploy DeepSeek V4 Pro locally with Ollama, llama.cpp or vLLM
The Ollama library has no official tag for it yet.
llama.cpp pulls the GGUF straight from Hugging Face; -c sets the context length.
llama-server -hf unsloth/DeepSeek-V4-Pro-0813-GGUF:Q4_K_M -c 8192
vLLM serves the original-precision weights, which needs far more memory than GGUF and usually more than one GPU.
vllm serve deepseek-ai/DeepSeek-V4-Pro-0813
File names follow that week's model-library snapshot; the definitions are in the methodology. Methodology
FAQ
How much VRAM does DeepSeek V4 Pro need?
At Q4_K_M with an 8K context DeepSeek V4 Pro needs about 894GB of VRAM; 1,076GB leaves comfortable headroom.
What is the cheapest GPU that runs DeepSeek V4 Pro?
No single card in the database fits DeepSeek V4 Pro; it needs multiple GPUs or a Mac Studio's unified memory.
Can you run DeepSeek V4 Pro with Ollama?
Not yet: the Ollama library has no official tag for it. You can load the GGUF with llama.cpp instead.