The issue is that Docling requires transformers >=4.46.0,<5.0.0 for VLM features, but you have transformers 5.3.0 installed. The error message is misleading—it says transformers isn't installed when the real problem is a version constraint violation.

To fix this, downgrade transformers:

pip install 'transformers>=4.46.0,<5.0.0'

Transformers 4.48.0 is recommended as it includes fixes for Qwen model compatibility.

Regarding why Qwen3-VL-2B-Instruct works—it's possible you're hitting different code paths, or the model successfully loads despite the version mismatch. However, for consistent behavior with VLM models, using the compatible transformers version range is recommended.

Also note: Qwen3.5-0.8B doesn't appear to be a VL (vision-language) model—it's a text-only model. For picture description in Docling, you'll need a vision-language model like Qwen3-VL variants.

To reply, just mention @dosu.

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}