ModelMergeQwenImage - ComfyUI Built-in Node Documentation

This node merges two Qwen image models by blending their individual components with adjustable weights.

This node merges two Qwen image models by blending their individual components with adjustable weights. Each weight controls how much the corresponding part of the second model contributes to the merged result, letting you mix position embeddings, text processing layers, image input layers, all 60 transformer blocks, and the output projection separately. The final result is a single MODEL that you can use wherever a normal model is expected.

Inputs

Parameter	Description	Data Type	Required	Range
model1	The first model to merge	MODEL	Yes	-
model2	The second model to merge	MODEL	Yes	-
pos_embeds.	Weight for positional embeddings blending (default: 1.0)	FLOAT	Yes	0.0 to 1.0 (step: 0.01)
img_in.	Weight for image input processing blending (default: 1.0)	FLOAT	Yes	0.0 to 1.0 (step: 0.01)
txt_norm.	Weight for text normalization blending (default: 1.0)	FLOAT	Yes	0.0 to 1.0 (step: 0.01)
txt_in.	Weight for text input processing blending (default: 1.0)	FLOAT	Yes	0.0 to 1.0 (step: 0.01)
time_text_embed.	Weight for time and text embedding blending (default: 1.0)	FLOAT	Yes	0.0 to 1.0 (step: 0.01)
transformer_blocks.0. to transformer_blocks.59.	Weight for each transformer block blending (default: 1.0). The node exposes one weight for each of the 60 transformer blocks.	FLOAT	Yes	0.0 to 1.0 (step: 0.01)
proj_out.	Weight for output projection blending (default: 1.0)	FLOAT	Yes	0.0 to 1.0 (step: 0.01)

Note: All weight inputs are required and share the same limits — a default of 1.0 with a valid range from 0.0 to 1.0, adjustable in steps of 0.01.

Outputs

Output Name	Description	Data Type
model	The merged model combining components from both input models with the specified weights	MODEL

This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub

---

Source fingerprint (SHA-256): 5f31f91f3d54d4c5085c684a98f64afd0a0f704693b6dd4f19bc35d3c5f74529