MultiModality [500] Internal Server Error - {"error":"POST predict: Post \"http://127.0.0.1:54956/completion\": EOF"}

[OnceCrazyer]

Bug description
I encountered this problem when using spring ai + ollama multimodal image recognition: [500] Internal Server Error - {"error": "POST predict: Post" http://127.0.0.1:53662/completion\ ": EOF"}
The model used is: joefamous/QVQ-72B-Preview: latest, but when I use the llava: 7b model, everything works fine

Environment
spring ai-M6
ollama
modal:joefamous/QVQ-72B-Preview: latest

Minimal Complete Reproducible example