Models / DeepSeek V4 Pro
Can I run DeepSeek V4 Pro locally?
DeepSeek · 1.60T parameters · mixture of experts, 49B active per token · up to 1024K context · license mit · Hugging Face
At Q4_K_M with a 32K context, DeepSeek V4 Pro needs about 1009 GB of GPU or unified memory by our estimate. The official checkpoint (FP8) is 865 GB on disk.
This model uses compressed/sparse attention; its KV cache is not modelled here (it is small compared to the weights).
Memory by quantization and context
| Quantization | Weights | Total at 8K | Total at 32K | Total at 128K |
|---|---|---|---|---|
| Q3_K_M (~4.0 bit) | 799 GB | 824 GB | 824 GB | 824 GB |
| Q4_K_M (~4.9 bit, popular default) | 978 GB | 1009 GB | 1009 GB | 1009 GB |
| Q5_K_M (~5.7 bit) | 1140 GB | 1175 GB | 1175 GB | 1175 GB |
| Q6_K (~6.6 bit) | 1312 GB | 1352 GB | 1352 GB | 1352 GB |
| Q8_0 (~8.5 bit, near-lossless) | 1699 GB | 1751 GB | 1751 GB | 1751 GB |
| F16 / BF16 (16 bit) | 3198 GB | 3295 GB | 3295 GB | 3295 GB |
Bits per weight from llama.cpp's quantize docs. F16 KV cache. Totals include our runtime-overhead estimate.
Cheapest hardware that fits (Q4_K_M, 32K)
No device or multi-GPU setup (up to 4 cards) in our catalog with a current price fits this. Try a lower quantization in the calculator below.
Runs in
| Tool | Platforms | How | Source |
|---|---|---|---|
| MLX (mlx-lm) Apple framework + CLI | macOS (Apple Silicon only) | MLX conversion: mlx-community/DeepSeek-V4-Pro | source |
Ollama: cloud only, it does not download the model to your machine.
Only tools with a public source are listed: an official build (ggml-org, lmstudio-community, mlx-community, bartowski or Unsloth on Hugging Face), the model card, ollama.com or the tool's own README. Apps built on llama.cpp, such as Jan and PocketPal AI, can usually load the same GGUF files. All local apps.
Try other settings
This model uses compressed/sparse attention; its KV cache is not modelled here (it is small compared to the weights).
Nothing in our catalog fits this setup
Cheapest real offer per device from AI Hardware Deals, converted to EUR at approximate rates. Multi-GPU rows multiply the card price. Try a smaller quantization, shorter context or more GPUs.
| Setup | Headroom |
|---|
Speed ceiling is memory bandwidth divided by the bytes of (active) weights read per token, shown only where we have a sourced bandwidth figure. Real decode speed is lower, especially with long contexts or when splitting across cards. All memory numbers are estimates; see the methodology.
FAQ
- How much VRAM does DeepSeek V4 Pro need?
- About 1009 GB at Q4_K_M with a 32K-token context (978 GB weights, KV cache not modelled, 30.3 GB overhead). These are estimates.
- Can I run DeepSeek V4 Pro on a single GPU?
- Not at Q4_K_M with 32K context on any single GPU in our catalog. Use several GPUs, a high-memory unified-memory machine, a smaller quantization or a shorter context.
- Which apps can run DeepSeek V4 Pro locally?
- Verified so far: MLX (mlx-lm). Each entry links to its source on this page.