Qwen3 Vl 30b

[tough topaz]

Currently, GGUF export for Qwen3-VL (including 30B) is not supported, as llama.cpp and Unsloth do not yet support saving Qwen3-VL models in GGUF format. You can fine-tune and run Qwen3-VL using Unsloth with safetensors, but GGUF export is unavailable for these vision-language models at this time. This limitation is explicitly mentioned in the documentation and confirmed in community discussions—users are advised to use transformers or vLLM for Qwen3-VL models instead of GGUF workflows.

If you need GGUF for Qwen3-VL 30B, you will have to wait for future updates from Unsloth and llama.cpp. Would you like more detail on current alternatives or updates on GGUF support for other Qwen3 models?

Sources:

• Qwen3-VL: Run & Fine-tune | Unsloth Documentation

💡 **Hint:** Mention @RunLLM in the thread for followups.

[ocean cloak]

I think, I'm not that sure, that the support was added a few minutes ago. @tough topaz