SiliconFlow icon

SiliconFlow

Interact with SiliconFlow AI models

Actions4

Vision → Analyze

AI-generated

Overview

This node interacts with the SiliconFlow AI platform to analyze images using a vision language model. It supports multiple image input formats including binary data, URLs, and base64 encoded images. Users provide a text prompt describing what they want to know about the images, and the node sends the images and prompt to the selected vision model for analysis. The node is useful for scenarios where automated image understanding or description is needed, such as content moderation, image tagging, or extracting information from images.

Use Case Examples

  1. Analyzing product images to generate descriptive tags for e-commerce listings.
  2. Processing user-uploaded photos to detect and describe objects or scenes.
  3. Extracting visual information from images for automated report generation.

Properties

Name Meaning
Model The vision language model to use for image analysis. Different models offer varying levels of performance and capabilities.
Images Collection of images to analyze. Supports multiple images (up to 9) with sources as binary data, URLs, or base64 encoded strings. Each image can specify its format and detail level for processing.
Prompt Text prompt describing what the user wants to know about the image(s). This guides the vision model's analysis.
Additional Fields Optional parameters to control the generation such as max tokens, temperature, top_p for diversity, and whether to stream the response.

Output

JSON

  • analysis - The textual analysis or description generated by the vision model for the provided images.
  • model - The model identifier used for the analysis.
  • finishReason - Reason why the model finished generating the response.
  • usage - Token usage statistics for the request.
  • imageCount - Number of images analyzed.
  • prompt - The prompt text used to guide the analysis.
  • _rawResponse - The full raw response from the SiliconFlow API for debugging or extended use.

Dependencies

  • SiliconFlow API key credential

Troubleshooting

  • Ensure that at least one image is provided in a valid format (binary, URL, or base64).
  • Verify that the base64 data is correctly formatted without the data:image prefix and contains valid base64 characters.
  • Check that the binary property specified contains valid image data and is accessible from the input.
  • If using URLs, ensure the URLs are reachable and point to valid image files.
  • Common error messages include missing image data, invalid base64 format, or no response from the vision model. These can be resolved by verifying input data and network connectivity.

Links

Discussion