Aliyun Bailian | 阿里云百炼Install

Summary

Analyze visual content by submitting images, videos, or mixed media to selected vision understanding models, and receive descriptive or interpretive answers based on the provided prompt.

Inputs

• Model | 模型 (required) — Select which vision understanding model to use, options include Qwen3-VL-Plus (advanced, suitable for complex scene understanding), Qwen-VL-Max (higher accuracy for complex analysis), and Qwen-VL-Plus (stable for everyday tasks).
• Media Type | 输入类型 (required) — Choose the type of visual input to provide: single or multiple images, a video URL, video frames as images, or a mix of images and video input.
• Prompt | 提问 (required) — Enter the question or prompt about the visual media to guide the model's analysis and response.
• Image URLs | 图片地址列表 — If input type includes images, provide one HTTP/HTTPS/OSS image URL per line.
• Video URL | 视频链接 — If input type includes a video URL, provide a single video URL.
• Video Frames | 视频帧列表 — If input type uses video frames, provide one image URL per frame line.
• FPS | 帧率 — Optional frames per second to specify for video inputs or mixed inputs containing video.
• Advanced Options | 高级选项 — Include optional advanced parameters such as system prompt (model instruction), maximum output token length, temperature (controls randomness), and nucleus sampling parameter (top P).

Output shape

a single JSON object containing the vision understanding model's textual interpretation or answer to the prompt, based on the input media; the exact output may vary by model and input type.

The output contains the model's response text, reflecting its analysis of the submitted images and/or videos. Inputs must be valid URLs accessible over HTTP/HTTPS or Alibaba OSS for media content. Model and media type choice affect processing and accuracy. No pagination or multiple records returned; one response per workflow execution.

Links

• n8n community nodes installation guide