DeepSeek R1 Distill Qwen 32B (Q3_K_M) — 15.5 GBon Apple M2 Pro 16GB
Overview
DeepSeek R1 Distill Qwen 32B is a 32.76B parameter dense language model by DeepSeek, with code, multilingual, thinking, tool-calls capabilities. It supports a context window of up to 131,072 tokens.
DeepSeek R1 Distill Qwen 32B is a 32.76-billion-parameter dense transformer from DeepSeek, distilled from the larger R1 reasoning model into a Qwen-based architecture. It excels at chain-of-thought reasoning, code generation, and multilingual tasks with built-in thinking capabilities. Compared to standard 30B-class instruct models, it provides stronger logical and mathematical reasoning. The model supports nine languages and a 128K context window, making it suitable for developers and researchers who need reasoning-focused inference on mid-range GPU setups.
At Q3_K_M quantization (low quality tier), the model weighs 14.84 GB. This fits within the 16 GB of VRAM on Apple M2 Pro 16GB, enabling full GPU inference.
Hardware Requirements
| Model size | 14.84 GB |
| VRAM available | 16 GB |
| VRAM used | 15.5 GB |
| System RAM | |
| Min RAM required | 0.7 GB |
| GPU layers | 61 / 64 |
| Context size | 512 |
| Backend | metal |
| Flash attention | Yes |
| Reading from disk | Yes |
Performance Notes
Deploy
Install llama.cpp
brew install llama.cppDownload Model
curl -L -o deepseek-r1-distill-qwen-32b.gguf "https://huggingface.co/unsloth/DeepSeek-R1-Distill-Qwen-32B-GGUF/resolve/main/DeepSeek-R1-Distill-Qwen-32B-Q3_K_M.gguf"Start Server
llama-server \
-m deepseek-r1-distill-qwen-32b.gguf \
--n-gpu-layers 61 \
--ctx-size 512 \
--flash-attnVerify
curl http://localhost:8080/healthFrequently Asked Questions
How much VRAM does DeepSeek R1 Distill Qwen 32B (Q3_K_M) need?
The Q3_K_M quantization of DeepSeek R1 Distill Qwen 32B requires 14.84 GB. 61 of 64 layers fit in the 16 GB of VRAM on Apple M2 Pro 16GB; remaining layers are offloaded to CPU.
Can I run DeepSeek R1 Distill Qwen 32B on Apple M2 Pro 16GB?
Yes, with reduced performance. Apple M2 Pro 16GB can run DeepSeek R1 Distill Qwen 32B (Q3_K_M), but only 61 of 64 layers fit in VRAM. The rest are offloaded to CPU.
What is quantization?
Quantization reduces a model's numerical precision from its original floating-point format to a more compact representation. This shrinks the file size and VRAM footprint, making it possible to run large models on consumer hardware. The trade-off is a small reduction in output quality. Q3_K_M compresses DeepSeek R1 Distill Qwen 32B from its original size down to 14.84 GB.
What quantization should I choose for DeepSeek R1 Distill Qwen 32B?
Q3_K_M is a low-quality quantization. Higher-quality quants (Q8, Q6) preserve more model accuracy but need more VRAM. Lower quants (Q4, Q3, Q2) reduce VRAM usage at the cost of some quality. Choose based on your available hardware and quality requirements.
What is flash attention and why is it enabled?
Flash attention is a memory-efficient algorithm that speeds up the attention mechanism in transformer models. It reduces VRAM usage during inference by avoiding the materialisation of the full attention matrix. For DeepSeek R1 Distill Qwen 32B on Apple M2 Pro 16GB, flash attention is enabled to maximise context length and throughput within the available 16 GB of VRAM.
Why are some layers offloaded to CPU?
Apple M2 Pro 16GB has 16 GB of VRAM, but DeepSeek R1 Distill Qwen 32B (Q3_K_M) requires approximately 14.84 GB. Only 61 of 64 layers fit in VRAM; the remaining layers run on CPU, which is slower but still functional.
How do I run DeepSeek R1 Distill Qwen 32B (Q3_K_M) with Ollama?
Run ollama run deepseek-r1:32b-qwen-distill-q3_k_m to start DeepSeek R1 Distill Qwen 32B (Q3_K_M). Ollama handles downloading the model weights automatically on first run.