Can you run Deepseek V3 locally?
Deepseek family · · 684.5B parameters (37B activated per token) · Hugging Face model card
Deepseek V3 is a mixture-of-experts model with 37.0B active parameters per token and 684.5B total open-weight model with a 163,840-token context window, usable locally with roughly 512 GB of memory at Q4-class quantization (calculated; measured speeds vary by hardware).
Minimum: 512 GB+ memory (Q4_K_M, calculated) · Recommended: 768 GB+ memory for full context (Q4_K_M, calculated)
How much VRAM does Deepseek V3 need at each quantization?
Deepseek V3 needs 451.8 GB of VRAM at Q4_K_M. The table below lists weights-only size and total VRAM including overhead for each common quantization level with 4k and 32k token contexts.
| Quantization | Bits / weight | Weights only | Total + KV @4k ctx | Total + KV @32k ctx |
|---|---|---|---|---|
| Q4_K_M | 4.8 | 410.7 GB | 451.8 GB | 451.8 GB |
| Q5_K_M | 5.7 | 487.7 GB | 536.5 GB | 536.5 GB |
| Q6_K | 6.6 | 564.7 GB | 621.2 GB | 621.2 GB |
| Q8_0 | 8.5 | 727.3 GB | 800.0 GB | 800.0 GB |
| FP16 | 16 | 1,369.1 GB | 1,506.0 GB | 1,506.0 GB |
Method: weights = parameters × bits-per-weight (Q4_K_M ≈ 4.8, Q5_K_M ≈ 5.7, Q6_K ≈ 6.6, Q8_0 ≈ 8.5, FP16 = 16), plus 10% loading overhead, plus KV cache from the verified architecture config (61 layers, 128 KV heads, head dim). Source: HF-verified 2026-09-12 via api (params=684.53B ctx=163840 license=?); VRAM figures are documented calculations, not measurements
Which GPUs can run Deepseek V3 locally?
At Q4_K_M with a 4k context, Deepseek V3 needs 451.8 GB of VRAM. The lists below are computed live from our GPU database and grouped by how much headroom the card has. Disclosure: CompareAIHardware.com participates in the Amazon Associates program and earns from qualifying purchases through links on this page. Affiliate relationships do not influence our recommendations.
Needs 2+ GPUs to run Deepseek V3
One of these cards is too small on its own, but a pair (tensor or pipeline parallel, ~90% efficiency) covers the 451.8 GB requirement. See our multi-GPU guide for setup.
| GPU | VRAM | Bandwidth | Type | |
|---|---|---|---|---|
| AMD Instinct MI325X | 256 GB ×2 | 6000 GB/s | Data Center GPU | |
| AMD Instinct MI355X | 288 GB ×2 | 8000 GB/s | Data Center GPU | |
| NVIDIA B300 (Blackwell Ultra) | 288 GB ×2 | 8000 GB/s | Data Center GPU |