ComfyUI-QwenVL
ComfyUI nodes for QwenVL (Qwen2.5-VL / Qwen3-VL): image caption, OCR, boxes, video summary.
Nodes (9)
Ask an image anything, locally — no API key, no Gemini bill
The QwenVL node, but with every knob — for when defaults aren't cutting it
Run a QwenVL captioner as a GGUF file — the low-VRAM path
The GGUF QwenVL node with llama.cpp's whole control panel
The friendlier QwenVL transformer node — tooltips included
QwenVL (HF Alt) with the sampling dials exposed — tuned captions, one widget at a time
Turn 'a cool knight' into a real prompt — with a tiny Qwen you already installed
Three photos in, one outfit-swap prompt out
Save a caption to a file without breaking your graph — the pack's humble text writer
ComfyUI-QwenVL
ComfyUI nodes for QwenVL models. Image caption, OCR, object boxes, and video summary. Works with Qwen2.5-VL and Qwen3-VL. 4-bit and 8-bit support.