DeepSeek R1 — local AI model by DeepSeek
Written by Jakub Rusinowski · Last updated
DeepSeek R1 rivals proprietary top-tier models in reasoning and coding tasks using a Mixture-of-Experts architecture. Known for its exceptional logic and math capabilities. Previous generation — superseded by DeepSeek V4, which folds R1's reasoning strengths into a newer MoE architecture. Still widely deployed and supported.
Variants
The smallest DeepSeek R1 variant needs about 6 GB of VRAM at Q4_K_M — quantized weights plus framework overhead, before any KV cache.
| Model | VRAM at Q4 | VRAM | Context | Run it |
|---|---|---|---|---|
| DeepSeek R1 Distill Llama 8B → 8B | ~5.6 GB | 128,000 | ollama run deepseek-r1:8b | |
| DeepSeek R1 Distill Qwen 32B → 32B | ~20.1 GB | 128,000 | ollama run deepseek-r1:32b | |
| DeepSeek R1 Distill Qwen 14B → 14B | ~9.3 GB | 128,000 | ollama run deepseek-r1:14b | |
| DeepSeek R1 (671B) → 671B | ~405.9 GB | 128,000 | ollama run deepseek-r1:671b |
Memory is quantized weights plus overhead at Q4_K_M, from the same engine as the GPU & VRAM checker.
How to run DeepSeek R1 locally
Install Ollama, then pull the tag.
ollama run deepseek-r1:8bPick a size above for its own VRAM figure, speed estimate and install command.
Licence
Commercial use permitted. No usage restrictions beyond attribution.
Applies to: DeepSeek R1 Distill Llama 8B, DeepSeek R1 Distill Qwen 32B, DeepSeek R1 Distill Qwen 14B, DeepSeek R1 (671B)Recommended GPU
The cheapest catalogued GPU that runs DeepSeek R1 locally (min 6 GB VRAM) is the Intel Arc B570 (10 GB).
Can I run DeepSeek R1 on my GPU?
- DeepSeek-R1-Distill-Qwen-14B on AMD Radeon RX 7800 XT
- DeepSeek-R1-Distill-Qwen-14B on AMD Radeon RX 7900 XT
- DeepSeek-R1-Distill-Qwen-32B on AMD Radeon RX 7900 XTX
- DeepSeek-R1-Distill-Qwen-14B on AMD Radeon RX 9060 XT 16GB
- DeepSeek-R1-Distill-Llama-8B on AMD Radeon RX 9060 XT 8GB
- DeepSeek-R1-Distill-Qwen-14B on AMD Radeon RX 9070
- DeepSeek-R1-Distill-Qwen-14B on AMD Radeon RX 9070 XT
- DeepSeek-R1-Distill-Qwen-14B on Apple M1
- DeepSeek-R1-Distill-Qwen-32B on Apple M1 Pro
- DeepSeek-R1-Distill-Qwen-32B on Apple M2 Pro
- DeepSeek-R1-Distill-Qwen-32B on Apple M3 Pro
- DeepSeek-R1-Distill-Qwen-32B on Apple M4
- DeepSeek-R1-Distill-Qwen-32B on Apple M5
- DeepSeek-R1-Distill-Llama-8B on Intel Arc B570
- DeepSeek-R1-Distill-Qwen-14B on Intel Arc B580
- DeepSeek-R1-Distill-Qwen-14B on NVIDIA GeForce RTX 3060 (12GB)
- DeepSeek-R1-Distill-Llama-8B on NVIDIA GeForce RTX 3070
- DeepSeek-R1-Distill-Llama-8B on NVIDIA GeForce RTX 3070 Ti
- DeepSeek-R1-Distill-Llama-8B on NVIDIA GeForce RTX 3080 (10GB)
- DeepSeek-R1-Distill-Qwen-32B on NVIDIA GeForce RTX 3090
- DeepSeek-R1-Distill-Llama-8B on NVIDIA GeForce RTX 4060
- DeepSeek-R1-Distill-Qwen-14B on NVIDIA GeForce RTX 4060 Ti 16GB
- DeepSeek-R1-Distill-Qwen-14B on NVIDIA GeForce RTX 4070
- DeepSeek-R1-Distill-Qwen-14B on NVIDIA GeForce RTX 4070 Super
- DeepSeek-R1-Distill-Qwen-14B on NVIDIA GeForce RTX 4070 Ti
- DeepSeek-R1-Distill-Qwen-14B on NVIDIA GeForce RTX 4070 Ti Super
- DeepSeek-R1-Distill-Qwen-14B on NVIDIA GeForce RTX 4080
- DeepSeek-R1-Distill-Qwen-14B on NVIDIA GeForce RTX 4080 Super
- DeepSeek-R1-Distill-Qwen-32B on NVIDIA GeForce RTX 4090
- DeepSeek-R1-Distill-Llama-8B on NVIDIA GeForce RTX 5060
- DeepSeek-R1-Distill-Qwen-14B on NVIDIA GeForce RTX 5060 Ti 16GB
- DeepSeek-R1-Distill-Llama-8B on NVIDIA GeForce RTX 5060 Ti 8GB
- DeepSeek-R1-Distill-Qwen-14B on NVIDIA GeForce RTX 5070
- DeepSeek-R1-Distill-Qwen-14B on NVIDIA GeForce RTX 5070 Ti
- DeepSeek-R1-Distill-Qwen-14B on NVIDIA GeForce RTX 5080
- DeepSeek-R1-Distill-Qwen-32B on NVIDIA GeForce RTX 5090
- DeepSeek-R1-Distill-Qwen-32B on NVIDIA L40S
- DeepSeek-R1-Distill-Qwen-32B on NVIDIA RTX 6000 Ada Generation
DeepSeek R1 — frequently asked questions
How much VRAM does DeepSeek R1 need?
DeepSeek R1 needs about 6 GB VRAM at Q4_K_M quantization for its smallest variant. Variants: DeepSeek R1 Distill Llama 8B (6 GB, Q4_K_M); DeepSeek R1 Distill Qwen 32B (20 GB, Q4_K_M); DeepSeek R1 Distill Qwen 14B (9 GB, Q4_K_M); DeepSeek R1 (671B) (406 GB, Q4_K_M). On Apple Silicon, unified memory counts toward this requirement.
Can I run DeepSeek R1 on an RTX 4090 (24 GB)?
Yes — DeepSeek R1 runs on an RTX 4090 (24 GB) and other 24 GB cards such as the RTX 3090. Smaller variants also fit comfortably on 8–16 GB GPUs at Q4_K_M.
What quantization should I use for DeepSeek R1?
Q4_K_M is the best balance of quality and VRAM for DeepSeek R1 in most cases. Choose Q8_0 for near-lossless quality if you have spare VRAM, or smaller quants (Q3/Q2) only when memory is tight.
How do I run DeepSeek R1 with Ollama?
Install Ollama, then run: ollama run deepseek-r1:8b. This downloads DeepSeek R1 and starts a local, OpenAI-compatible endpoint — no internet connection is needed after the initial download.