What GPU do you need to run DeepSeek V4 Pro?
1600B params 1M max context Tool useReasoning
DeepSeek V4 Pro doesn't fit on any single GPU in our database. Try multiple GPUs or system RAM offloading in the full tool.
MMLU 87.5 SWE-bench Verified 80.6
VRAM needed per quantization
| Quantization | Weights | 16K context | 32K context | Full 1M |
|---|---|---|---|---|
| Q4_K_M | 900.0 GB | 901.0 GB | 902.1 GB | 966.6 GB |
| Q8_0 | 1670.0 GB | 1671.0 GB | 1672.1 GB | 1736.6 GB |
Weights plus fp16 KV cache. Leave ~1 GB headroom for the inference engine.
GPUs that can run DeepSeek V4 Pro
None on its own — open the tool to try multiple GPUs or system RAM offloading.