To fine-tune Qwen3, update via:

pip install --upgrade --force-reinstall --no-cache-dir unsloth unsloth_zoo

Qwen3 MOE models

FOR MOE models PLEASE USE FastModel and not FastLanguageModel

GitHub release out now! MOE models also supported. Qwen3-30B-A3B fits on 17.5GB VRAM: https://github.com/unslothai/unsloth/releases/tag/May-2025

GRPO

UPDATE May 13: Qwen3 GRPO now works with the latest vLLM version. Notebook

Fixes

A huge thank you to @Datta0 for this PR!! #2211

UPDATE May 13: We fixed an issue affecting inference for both instruct and base models