---
base_model: Qwen/Qwen3-14B
---

[EXL3](https://github.com/turboderp-org/exllamav3) quantization of [Qwen3-14B](https://huggingface.co/Qwen/Qwen3-14B), 6 bits per weight.

## HumanEval (argmax)

| Model                                                                             | Q4   | Q6   | Q8   | FP16 |
| --------------------------------------------------------------------------------- | ---- | ---- | ---- | ---- |
| [Qwen3-14B-exl3-4bpw](https://huggingface.co/isogen/Qwen3-14B-exl3-4bpw)          | 88.4 | 89.0 | 89.0 | 89.0 |
| [Qwen3-14B-exl3-6bpw](https://huggingface.co/isogen/Qwen3-14B-exl3-6bpw)          | 89.6 | 88.4 | 89.6 | 89.0 |
| [Qwen3-8B-exl3-4bpw](https://huggingface.co/turboderp/Qwen3-8B-exl3)              | 86.0 | 85.4 | 86.0 | 87.2 |
| [Qwen3-8B-exl3-6bpw](https://huggingface.co/turboderp/Qwen3-8B-exl3)              | 84.8 | 86.0 | 87.2 | 87.2 |
| [Qwen3-8B-exl3-8bpw-h8](https://huggingface.co/turboderp/Qwen3-8B-exl3)           | 86.0 | 87.2 | 86.6 | 86.6 |
| [Qwen3-30B-A3B-exl3-2.25bpw](https://huggingface.co/turboderp/Qwen3-30B-A3B-exl3) |      |      |      | 88.4 |
| [Qwen3-30B-A3B-exl3-3bpw](https://huggingface.co/turboderp/Qwen3-30B-A3B-exl3)    |      |      |      | 89.6 |
| [Qwen3-30B-A3B-exl3-4bpw](https://huggingface.co/turboderp/Qwen3-30B-A3B-exl3)    |      |      |      | 92.1 |
| [Qwen3-32B-exl3-4bpw](https://huggingface.co/turboderp/Qwen3-32B-exl3)            |      |      |      | 91.5 |