DeepSeek R1 Models — Hardware Requirements

13 DeepSeek R1 models from DeepSeek and the community, from the smallest that runs in 0.7 GB of VRAM up to 684.5B parameters. Every row links to full quantization tables and GPU compatibility.

All DeepSeek R1 Models by Size

ModelParamsContext
DeepSeek R1 Distill Qwen 1.5B1.5B
DeepSeek R1 Distill Qwen 1.5B1.8B131K
DeepSeek R1 Distill Qwen 7B7.6B131K
DeepSeek R1 Distill Llama 8B8.0B131K
DeepSeek R1 0528 Qwen3 8B8.2B131K
DeepSeek R1 Distill Qwen 14B14.8B131K
DeepSeek R1 Distill Qwen 14B Abliterated v214.8B131K
DeepSeek R1 Distill Qwen 32B32.8B131K
DeepSeek R1 Distill Qwen 32B Abliterated32.8B131K
DeepSeek R1 Distill Llama 70B70.6B131K
DeepSeek R1 0528684.5B164K
DeepSeek R1684.5B164K
DeepSeek R1 Zero684.5B164K

How DeepSeek R1 Compares — Benchmark Rating

DeepSeek R1 0528 is the highest-rated DeepSeek R1 model with an overall benchmark rating of 55.9/100 — #23 among 75 open models. The top proprietary model, GPT 5.5, scores 88.8. Click a model to see its full benchmark breakdown.

GPT 5.5 · proprietary88.8
Claude Opus 4.7 · proprietary87.6
Claude Fable 5 · proprietary86.6
GPT 5.4 · proprietary86.6
Claude Opus 4.8 · proprietary84.4
Composite of normalized public benchmark scores (methodology) · DeepSeek R1 · other models

Frequently Asked Questions

How much VRAM do I need to run a DeepSeek R1 model?
The smallest DeepSeek R1 model, DeepSeek R1 Distill Qwen 1.5B, runs from 0.7 GB of VRAM at an aggressive quantization. Larger family members need proportionally more — see the table above for every model.
Which DeepSeek R1 models can I run on a 16 GB GPU?
9 of 13 DeepSeek R1 models fit in 16 GB of VRAM at some quantization, including DeepSeek R1 0528 Qwen3 8B, DeepSeek R1 Distill Qwen 14B, DeepSeek R1 Distill Qwen 1.5B.
What is the most popular DeepSeek R1 model to run locally?
DeepSeek R1 0528 Qwen3 8B is the most downloaded DeepSeek R1 model in local-friendly quantized formats. It runs from 2.9 GB of VRAM.
How do DeepSeek R1 models score on benchmarks?
DeepSeek R1 0528 leads the family with an overall benchmark rating of 55.9/100, ranking #23 among 75 open models, while the top proprietary model, GPT 5.5, scores 88.8. See the comparison chart above for the full standings.