I’m using @felixvemmer 's Vercel AI Node (Google Gemini) to submit files to Gemini 2 using its multimodal capabilities. Although I can get this to work for single files, I’ve not been able to submit an array of files to the model so that it analyses them in the round (as it does in Google AI Studio when testing); the best I’ve managed is for it to process the array of binaries in sequence, which isn’t what I’m after.
Can anybody provide some advice on how to submit an array of binaries to a multimodal model like Gemini for a batch analysis?
Many thanks!
Please share your workflow
Share the output returned by the last node
repeated output based on the first array item:
Feedback Response
The image shows a stack of colorful, ornate fabrics, possibly saris or other traditional garments. The top layer is a green fabric with silver floral patterns and sequin embellishments. Below that, there are layers of pink, gold, and blue fabrics, each with intricate embroidery and beadwork. The bottom layer has dangling beaded embellishments.
The image shows a stack of colorful, ornate fabrics, possibly saris or other traditional garments. The top layer is a green fabric with silver floral patterns and sequin embellishments. Below that, there are layers of pink, gold, and blue fabrics, each with intricate embroidery and beadwork. The bottom layer has dangling beaded embellishments.
Information on your n8n setup
- n8n version: 1.85.4
- Database (default: SQLite): SQLite
- Running n8n via (Docker, npm, n8n cloud, desktop app): Docker
- Operating system: Ubuntu 24
maybe you could merge several files into one file? 