When multiple LLM nodes are connected, image information is not passed to the next LLM node #5434
Description
Describe the bug
I discovered a bug where, after connecting multiple LLM nodes and a condition agent node and uploading an image, the system failed to extract the image context properly.
To identify when the issue occurs, I conducted a simple test using the following setup:
When uploading an image through Flowise’s internal chat, the first LLM node successfully extracted the text from the image.
However, the second LLM node failed to retrieve the image information.
When checking the result values of each flow, they are as follows:
LLM1 node:
LLM2 node:
In the llm1 node, the user field contains image information, but in the llm2 node, the user field does not include any image information.
The prompts for the llm1 node and llm2 node are as follows:
Extract the text content from the file uploaded by the user.
Question: {{question}}
Flowise Version
V 3.0.7