Qwen3 32B (Q2_K) — 19 GBon OVH l40s-1-gpu

Code Multilingual Thinking Tool Calls

Q2_K OVH l40s-1-gpu

Overview

Qwen3 32B is a 32B parameter dense language model by Qwen, with code, multilingual, thinking, tool-calls capabilities. It supports a context window of up to 40,960 tokens.

Qwen3 32B is a 32-billion-parameter dense transformer from Alibaba's Qwen team, combining thinking capabilities with strong code generation, tool calling, and multilingual support. It occupies a mid-range parameter class that balances reasoning depth with practical deployment requirements, outperforming many larger models on math and logic benchmarks. The model supports 14 languages including English, Chinese, and Arabic. With a 40K context window and flash attention, it fits on a single high-end GPU at Q4 quantization for self-hosted inference.

At Q2_K quantization (low quality tier), the model weighs 11.5 GB. This fits within the 48 GB of VRAM on OVH l40s-1-gpu, enabling full GPU inference.

The NVIDIA L40S is a datacenter GPU with 48 GB of GDDR6 VRAM and 864 GB/s memory bandwidth. It delivers 362 FP16 TFLOPS with Ada Lovelace architecture. A versatile GPU for AI inference, training, and graphics workloads. Handles quantized models up to 40B parameters comfortably.

Hardware Requirements

Model size	11.5 GB
VRAM available	48 GB
VRAM used	19 GB
System RAM	80 GB
Min RAM required	0 GB
GPU layers	64 / 64
Context size	40,960
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/qwen3-32b/q2_k/nvidia-l40s.yaml) apply

Generated `values.yaml`

/values/qwen3-32b/q2_k/nvidia-l40s.yaml

Loading values…

Install llama.cpp

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

Download Model

curl -L -o qwen3-32b.gguf "https://huggingface.co/unsloth/Qwen3-32B-GGUF/resolve/main/Qwen3-32B-Q2_K.gguf"

Start Server

llama-server \
  -m qwen3-32b.gguf \
  --n-gpu-layers 64 \
  --ctx-size 40960 \
  --flash-attn

Verify

curl http://localhost:8080/health

Frequently Asked Questions

How much VRAM does Qwen3 32B (Q2_K) need?

The Q2_K quantization of Qwen3 32B requires 11.5 GB. All 64 layers fit in the 48 GB of VRAM available on OVH l40s-1-gpu, enabling full GPU acceleration.

Can I run Qwen3 32B on OVH l40s-1-gpu?

Yes. OVH l40s-1-gpu provides 48 GB of VRAM, which is enough to run Qwen3 32B (Q2_K) 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 compresses Qwen3 32B from its original size down to 11.5 GB.

What quantization should I choose for Qwen3 32B?

Q2_K 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 Qwen3 32B on OVH l40s-1-gpu, flash attention is enabled to maximise context length and throughput within the available 48 GB of VRAM.

How do I run Qwen3 32B (Q2_K) with Ollama?

Run ollama run qwen3:32b-q2_k to start Qwen3 32B (Q2_K). Ollama handles downloading the model weights automatically on first run.

Last updated: March 5, 2026