DeepSeek · Mixture-of-experts · MIT
Can I run DeepSeek-R1 671B-A37B (MoE) locally?
The closest open weights to frontier reasoning. ~400 GB even at Q4 — a server or a maxed multi-GPU rig only.
- Parameters
- 671B (37B active)
- VRAM at Q4
- 424 GB
- Max context
- 131K
- Released
- 2025-01
Memory needed by quantisation
Total includes weights, an 8K-token KV cache and runtime overhead. Lower quants trade quality for size.
| Quant | Bits/weight | Weights | Total needed | Quality |
|---|---|---|---|---|
| FP16 | 16 | 1250 GB | 1331 GB | Full precision. Reference quality, 2× the size of 8-bit. |
| Q8_0 | 8.5 | 664 GB | 722 GB | Effectively lossless. The safe choice when it fits. |
| Q6_K | 6.56 | 512 GB | 564 GB | Near-lossless; quality loss is hard to measure. |
| Q5_K_M | 5.67 | 443 GB | 492 GB | Very good. A common sweet spot above Q4. |
| Q4_K_M | 4.83 | 377 GB | 424 GB | The default. Best size/quality trade-off for local use. |
| Q3_K_M | 3.91 | 305 GB | 349 GB | Noticeable degradation; useful to squeeze a size up. |
| Q2_K | 3.35 | 262 GB | 303 GB | Aggressive. Quality drops a lot — last resort to fit. |
Which hardware runs DeepSeek-R1 671B-A37B (MoE)?
Best quantisation that fits each device at 8K context, with a rough speed estimate. Try your exact setup →
| Hardware | Memory | Fit | Best quant | Speed |
|---|---|---|---|---|
| Mac Studio · M3 Ultra, 512 GB | 512 GB | Runs (tight) | Q3 | ~21 tok/s |
| RTX 3060 12 GB | 12 GB | Won't fit | — | — |
| RTX 4060 Ti 8 GB | 8 GB | Won't fit | — | — |
| RTX 4060 Ti 16 GB | 16 GB | Won't fit | — | — |
| RTX 3080 10 GB | 10 GB | Won't fit | — | — |
| RTX 5070 | 12 GB | Won't fit | — | — |
| RTX 4070 Super | 12 GB | Won't fit | — | — |
| Radeon RX 7900 XT | 20 GB | Won't fit | — | — |
| RTX 5070 Ti | 16 GB | Won't fit | — | — |
| RTX 4070 Ti Super | 16 GB | Won't fit | — | — |
| RTX 3090 | 24 GB | Won't fit | — | — |
| Radeon RX 7900 XTX | 24 GB | Won't fit | — | — |
| RTX 5080 | 16 GB | Won't fit | — | — |
| RTX 4080 Super | 16 GB | Won't fit | — | — |
| RTX 4090 | 24 GB | Won't fit | — | — |
| RTX 5090 | 32 GB | Won't fit | — | — |
| RTX A6000 | 48 GB | Won't fit | — | — |
| RTX 6000 Ada | 48 GB | Won't fit | — | — |
| A100 80 GB | 80 GB | Won't fit | — | — |
| H100 80 GB | 80 GB | Won't fit | — | — |
| Mac · M1/M2/M3 (base), 8 GB | 8 GB | Won't fit | — | — |
| Mac · M1/M2/M3 (base), 16 GB | 16 GB | Won't fit | — | — |
| Mac · M4 (base), 24 GB | 24 GB | Won't fit | — | — |
| Mac · M4 Pro, 48 GB | 48 GB | Won't fit | — | — |
| Mac · M1/M2/M3 Max, 32 GB | 32 GB | Won't fit | — | — |
| Mac · M4 Max, 64 GB | 64 GB | Won't fit | — | — |
| Mac · M1/M2/M3 Max, 64 GB | 64 GB | Won't fit | — | — |
| Mac · M3/M4 Max, 128 GB | 128 GB | Won't fit | — | — |
| Mac Studio · M1/M2 Ultra, 128 GB | 128 GB | Won't fit | — | — |
| Mac Studio · M3 Ultra, 256 GB | 256 GB | Won't fit | — | — |
| CPU only · 8 GB RAM | 8 GB | Won't fit | — | — |
| CPU only · 16 GB RAM | 16 GB | Won't fit | — | — |
| CPU only · 32 GB RAM | 32 GB | Won't fit | — | — |
| CPU only · 64 GB RAM | 64 GB | Won't fit | — | — |
| CPU only · 128 GB RAM | 128 GB | Won't fit | — | — |
What DeepSeek-R1 671B-A37B (MoE) is good for
Related models
FAQ
How much VRAM does DeepSeek-R1 671B-A37B (MoE) need?
At Q4_K_M, DeepSeek-R1 671B-A37B (MoE) needs about 424 GB including a 8K-token context and overhead (377 GB for the weights alone). Higher quantisation needs more; see the table for every level.
What is the cheapest way to run DeepSeek-R1 671B-A37B (MoE)?
DeepSeek-R1 671B-A37B (MoE) is large — it needs a high-memory workstation GPU, multi-GPU setup, or a high-RAM Apple-silicon Mac. See the hardware table below.
Is DeepSeek-R1 671B-A37B (MoE) good for reasoning & math?
The closest open weights to frontier reasoning. ~400 GB even at Q4 — a server or a maxed multi-GPU rig only.
Estimates — see how we compute these. Memory figures assume an 8K context; long-context use needs more.