Model family · 5 sizes
DeepSeek-R1-Distill: which size runs locally?
DeepSeek-R1-Distill comes in 5 sizes, from 7B to 32B. Bigger sizes are more capable but need proportionally more memory. Here is each size with its Q4_K_M weight, the memory it needs, and the hardware that runs it.
- Sizes
- 5
- Smallest
- 7B
- Largest
- 32B
- Runs from
- 16GB
The DeepSeek-R1-Distill lineup
- DeepSeek-R1-Distill-Qwen 7B7B · ~4.68 GB Q4_K_M · needs ~6 GB
- DeepSeek-R1-Distill-Llama 8B8B · ~4.92 GB Q4_K_M · needs ~7 GB
- DeepSeek-R1-0528-Qwen3-8B8.19B · ~4.68 GB Q4_K_M
- DeepSeek-R1-Distill-Qwen 14B14B · ~8.99 GB Q4_K_M · needs ~11 GB
- DeepSeek-R1-Distill-Qwen 32B32B · ~19.85 GB Q4_K_M · needs ~22 GB
"Needs" is the sourced minimum memory for Q4_K_M with a small context. Larger context needs more.
Which DeepSeek-R1-Distill fits your memory
No DeepSeek-R1-Distill size fits 8GB; even DeepSeek-R1-Distill-Qwen 7B needs more.
Largest that fits: DeepSeek-R1-Distill-Qwen 14B (14B), best case on Nvidia GeForce RTX 4080 (16GB).
Largest that fits: DeepSeek-R1-Distill-Qwen 32B (32B), best case on Nvidia GeForce RTX 4090 (24GB). Comfortable up to DeepSeek-R1-Distill-Qwen 14B (14B).
Largest that fits: DeepSeek-R1-Distill-Qwen 32B (32B), best case on Nvidia GeForce RTX 5090 (32GB).
Largest that fits: DeepSeek-R1-Distill-Qwen 32B (32B), best case on Apple M5 Pro (48GB).
Largest that fits: DeepSeek-R1-Distill-Qwen 32B (32B), best case on Apple M4 Max (64GB).
Largest that fits: DeepSeek-R1-Distill-Qwen 32B (32B), best case on Apple M5 Max (128GB).
Largest that fits: DeepSeek-R1-Distill-Qwen 32B (32B), best case on Apple M3 Ultra (256GB).
Best case means the most capable device at that size (usually a discrete GPU). A Mac at the same size sits roughly one rung lower; see the per-size breakdown on each memory budget page.
FAQ
Which DeepSeek-R1-Distill size should I run locally?
Pick the largest size your memory allows. On 16GB (best case) up to DeepSeek-R1-Distill-Qwen 14B; On 24GB (best case) up to DeepSeek-R1-Distill-Qwen 32B; On 32GB (best case) up to DeepSeek-R1-Distill-Qwen 32B; On 48GB (best case) up to DeepSeek-R1-Distill-Qwen 32B; On 64GB (best case) up to DeepSeek-R1-Distill-Qwen 32B; On 128GB (best case) up to DeepSeek-R1-Distill-Qwen 32B; On 256GB (best case) up to DeepSeek-R1-Distill-Qwen 32B. Smaller sizes run faster and leave headroom for context.
What is the smallest DeepSeek-R1-Distill model?
DeepSeek-R1-Distill-Qwen 7B at 7B parameters, about 4.68 GB on disk at Q4_K_M and roughly 6 GB of memory to run. It is the one to use on phones and 8 GB machines.
What is the largest DeepSeek-R1-Distill model and what does it need?
DeepSeek-R1-Distill-Qwen 32B at 32B, about 19.85 GB at Q4_K_M and roughly 22 GB of memory. It fits a high-memory desktop GPU or Mac.
Understand the numbers
Short guides to the ideas behind DeepSeek-R1-Distill's memory and quant figures.
Sources
- github.com
- huggingface.co/bartowski/DeepSeek-R1-Distill-Llama-8B-GGUF
- huggingface.co/bartowski/DeepSeek-R1-Distill-Qwen-14B-GGUF
- huggingface.co/bartowski/DeepSeek-R1-Distill-Qwen-32B-GGUF
- huggingface.co/bartowski/DeepSeek-R1-Distill-Qwen-7B-GGUF
- huggingface.co/deepseek-ai/DeepSeek-R1-0528-Qwen3-8B
- huggingface.co/deepseek-ai/DeepSeek-R1-0528-Qwen3-8B/blob
- huggingface.co/unsloth
- ollama.com/library/deepseek-r1
- ollama.com/library/deepseek-r1:8b-0528-qwen3-q4_K_M
Memory figures are estimates at Q4_K_M. See methodology.