Hello,Whats the best way to get multimodal image

Hello,
Whats the best way to get multimodal image inspection working with anthropic models & litellm?

The integration works perfectly with most openai models but when using anthropic I get this.