ComfyUI Node
ModelScope VLM
Analyze images with text using ModelScope Vision-Language Models.
ModelScope VLM
- response
◄model_idQwen/QVQ-72B-Preview►
◄image_url—►
◄prompt—►
◄system_prompt—►
◄api_key—►
◄request_timeout120►
CategoryModelScope
Inputs (6)
| Name | Type | Default | Description |
|---|---|---|---|
| model_id | STRING | Qwen/QVQ-72B-Preview | ModelScope VLM model ID. |
| image_url | STRING | URL of the image to analyze. Must be publicly accessible. | |
| prompt | STRING | Question or instruction about the image. | |
| system_promptopt | STRING | Optional system prompt to guide the assistant's behavior. | |
| api_keyopt | STRING | ModelScope API key. If empty, reads MODELSCOPE_API_KEY or MODELSCOPE_ACCESS_TOKEN from environment. | |
| request_timeoutopt | INT | 12030–300 | Request timeout in seconds. VLM inference can be slow, recommend 120+ seconds. |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| response | STRING | VLM response describing or answering about the image. |