DeepSeek R1
DeepSeek · MIT
Mixture of Experts: 671B total parameters, 37B active per token.
DeepSeek's reasoning-focused model. Same MoE architecture as V3 (671B total, ~37B active per token). Excels at chain-of-thought reasoning, math, and complex code. Requires the same massive hardware as V3: multi-GPU or 128GB+ Apple Silicon at heavy quantization. For most home users, the distilled versions (7B, 32B) are more practical.
- Parameters
- 671B
- Architecture
- MoE (37B active)
- Context
- 65,536 tokens
- Released
- 2025-01-20
- Engines
- llama.cpp, vLLM, SGLang
Parameters
671B
VRAM
360 GB
Context
64K
Formats
4
GPUs
13
DeepSeek R1 (671B) requires 360 GB VRAM at recommended quality (FP16). On NVIDIA Grace Blackwell Ultra GB300, expect approximately 20 tok/s at Q4_K_M.
Source: OwnRig methodology
360 GB
FP16
335 GB
64K tokens
Reasoning
VRAM Requirements
| Quality | Quantization | VRAM | File Size |
|---|---|---|---|
| full | FP16 | 360 GB | 335 GB |
| efficient | Q4_K_M | 180 GB | 168 GB |
| compressed | Q3_K_M | 145 GB | 135 GB |
| compressed | Q2_K | 115 GB | 108 GB |
Context Length Impact
KV cache VRAM at FP16 quality. Longer context = more memory.
| Context | KV Cache | Total VRAM |
|---|---|---|
| 2K | 1 GB | 361 GBexceeds 24 GB |
| 4K | 2 GB | 362 GBexceeds 24 GB |
| 8K | 4.1 GB | 364.1 GBexceeds 24 GB |
| 16K | 8.2 GB | 368.2 GBexceeds 24 GB |
| 32K | 16.4 GB | 376.4 GBexceeds 24 GB |
| 64K | 32.8 GB | 392.8 GBexceeds 24 GB |
Compatible GPUs
13 devices| NVIDIA Grace Blackwell Ultra GB300 | Q4_K_M | 20 tok/s | Good |
| Apple M4 Max (128GB Unified) | Q2_K | 4 tok/s | Marginal |
| Apple M4 Ultra (192GB) | Q2_K | 6 tok/s | Marginal |
| NVIDIA GeForce RTX 4090 | Q2_K | 1 tok/s | Not viable |
| NVIDIA GeForce RTX 5090 | Q2_K | 1 tok/s | Not viable |
| AMD Radeon RX 7600 | Q2_K | – | Not viable |
| AMD Radeon RX 7900 XTX | Q2_K | – | Not viable |
| AMD Radeon Pro W7900 | Q2_K | – | Not viable |
| NVIDIA RTX PRO 6000 Blackwell | Q2_K | – | Not viable |
| NVIDIA RTX PRO 6000 Blackwell Max-Q | Q2_K | – | Not viable |
| AMD Radeon RX 9070 | Q2_K | – | Not viable |
| AMD Radeon RX 9060 XT 16GB | Q2_K | – | Not viable |
| AMD Radeon RX 9060 XT 8GB | Q2_K | – | Not viable |
Showing 13 of 13 entries
Frequently Asked Questions
- How much VRAM does DeepSeek R1 need?
- DeepSeek R1 requires 360 GB VRAM at recommended quality (FP16). At lower quality settings, it can fit in as little as 115 GB.
- What is the best GPU for DeepSeek R1?
- The NVIDIA Grace Blackwell Ultra GB300 delivers the best performance for DeepSeek R1, achieving 20 tok/s at Q4_K_M with an good rating.
- What quantization should I use for DeepSeek R1?
- For the best quality, use FP16 (360 GB VRAM). If your GPU has limited VRAM, Q2_K (115 GB) is the most efficient option with acceptable quality.
- Is DeepSeek R1 good for coding?
- DeepSeek R1 supports coding use cases. For the best coding experience, pair it with an embedding model for local RAG.
Data confidence: estimated. Source
VRAM requirements are calculated from model parameters and may vary by inference engine, context length, and batch size. Performance estimates are based on community benchmarks and should be verified for your specific configuration.DeepSeek is a trademark of its respective owner. OwnRig is not affiliated with or endorsed by the model creator.