DeepSeek R1 Distill Qwen 32B (Q2_K_L) — 32.9 GBon NVIDIA L40S 192GB

Code Multilingual Thinking Tool Calls

Q2_K_L NVIDIA L40S 192GB

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 Q2_K_L quantization (low quality tier), the model weighs 11.64 GB. This fits within the 192 GB of VRAM on NVIDIA L40S 192GB, enabling full GPU inference.

Hardware Requirements

Model size	11.64 GB
VRAM available	192 GB
VRAM used	32.9 GB
System RAM
Min RAM required	0 GB
GPU layers	64 / 64
Context size	131,072
Backend	cuda13
Flash attention	Yes
Reading from disk	Yes

Deploy

Prerequisites

Ensure your GPU nodes are prepared with the NVIDIA container toolkit:

ansible-playbook prositronic.infra.nvidia_container_toolkit

Command

helmfile --state-values-file <(curl -s https://www.prositronic.eu/values/deepseek-r1-distill-qwen-32b/q2_k_l/nvidia-l40s-192gb.yaml) apply

Generated `values.yaml`

/values/deepseek-r1-distill-qwen-32b/q2_k_l/nvidia-l40s-192gb.yaml

Loading values…

Install llama.cpp

Install llama.cpp from the official build instructions for your platform and backend.

Download 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-Q2_K_L.gguf"

Start Server

llama-server \
  -m deepseek-r1-distill-qwen-32b.gguf \
  --n-gpu-layers 64 \
  --ctx-size 131072 \
  --flash-attn

Verify

curl http://localhost:8080/health

Frequently Asked Questions

How much VRAM does DeepSeek R1 Distill Qwen 32B (Q2_K_L) need?

The Q2_K_L quantization of DeepSeek R1 Distill Qwen 32B requires 11.64 GB. All 64 layers fit in the 192 GB of VRAM available on NVIDIA L40S 192GB, enabling full GPU acceleration.

Can I run DeepSeek R1 Distill Qwen 32B on NVIDIA L40S 192GB?

Yes. NVIDIA L40S 192GB provides 192 GB of VRAM, which is enough to run DeepSeek R1 Distill Qwen 32B (Q2_K_L) with all layers on the GPU for optimal performance.

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. Q2_K_L compresses DeepSeek R1 Distill Qwen 32B from its original size down to 11.64 GB.

What quantization should I choose for DeepSeek R1 Distill Qwen 32B?

Q2_K_L 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 NVIDIA L40S 192GB, flash attention is enabled to maximise context length and throughput within the available 192 GB of VRAM.

How do I run DeepSeek R1 Distill Qwen 32B (Q2_K_L) with Ollama?

Run ollama run deepseek-r1:32b-qwen-distill-q2_k_l to start DeepSeek R1 Distill Qwen 32B (Q2_K_L). Ollama handles downloading the model weights automatically on first run.

Last updated: March 5, 2026