# Qwen2.5-Coder-32B-Instruct GPU requirements

Qwen2.5-Coder-32B-Instruct has 33B parameters. To serve it, plan for about 79 GB of GPU memory in 16-bit, 39 GB in 8-bit or 20 GB in 4-bit, including about 20% for the KV cache and activations.

[Deploy it on Arrakis](/console/#/models?model=Qwen/Qwen2.5-Coder-32B-Instruct).

## What it takes to serve it

| Precision | Memory and the smallest setup |
|---|---|
| 16-bit | 79 GB: one [H100](/gpus/h100.html.md) or [A100](/gpus/a100.html.md), H100 from $1.65 per GPU-hour |
| 8-bit | 39 GB: one [L40S](/gpus/l40s.html.md), [RTX 6000 Ada](/gpus/rtx-6000-ada.html.md) or [RTX A6000](/gpus/rtx-a6000.html.md), L40S from $0.44 per GPU-hour |
| 4-bit | 20 GB: one [RTX 4090](/gpus/rtx-4090.html.md) or [RTX 3090](/gpus/rtx-3090.html.md), RTX 3090 from $0.19 per GPU-hour |

## What it takes to fine-tune it

| Method | Memory and the smallest setup |
|---|---|
| QLoRA | 23 GB: one [RTX 4090](/gpus/rtx-4090.html.md) or [RTX 3090](/gpus/rtx-3090.html.md), RTX 3090 from $0.19 per GPU-hour |
| LoRA | 79 GB: one [H100](/gpus/h100.html.md) or [A100](/gpus/a100.html.md), H100 from $1.65 per GPU-hour |
| Full fine-tuning | 524 GB: 4 × [H200](/gpus/h200.html.md), from $16.12 per hour for all 4 |

## About the model

| | |
|---|---|
| Parameters | `33B` |
| Context window (tokens) | `32,768` |
| Published in | `BF16` |
| License | `apache-2.0` |
| Released | `2024-11-06` |
| Weights | [Qwen/Qwen2.5-Coder-32B-Instruct](https://huggingface.co/Qwen/Qwen2.5-Coder-32B-Instruct) |

## Serve it yourself

```
vllm serve Qwen/Qwen2.5-Coder-32B-Instruct --tensor-parallel-size 1
```

## Questions

**How much GPU memory does Qwen2.5-Coder-32B-Instruct need?** To serve it, about 79 GB in 16-bit, 39 GB in 8-bit, 20 GB in 4-bit, including about 20% for the KV cache and activations. To fine-tune it, about 23 GB with QLoRA, 79 GB with LoRA and 524 GB for full fine-tuning with Adam, before activations.

**Can I run Qwen2.5-Coder-32B-Instruct on one GPU?** Yes. In 16-bit it needs about 79 GB, which fits on one H100 or A100.

**What does it cost to run Qwen2.5-Coder-32B-Instruct?** In 16-bit it needs one H100 or A100; in 4-bit, one RTX 4090 or RTX 3090. Each setup on this page shows the lowest price per GPU-hour on Arrakis right now, and you pay by the hour from a prepaid balance.

**How do I deploy Qwen2.5-Coder-32B-Instruct on Arrakis?** Open Models in the Arrakis console, paste Qwen/Qwen2.5-Coder-32B-Instruct, choose vLLM, SGLang or llama.cpp, and rent a GPU it fits on. You get an OpenAI-compatible endpoint with its own API key, and root SSH to the machine.


More models: [/models](/models.html.md).