Prompt cache for Claude Sonnet used via Vertex AI

[david-sabata]

Issue

Hi!
Is the prompt caching support hardcoded to work only for direct Anthropic API?

I'm using Claude via Vertex AI and the prompt cache label isn't showing up even though Vertex does support prompt caching for Anthropic models: https://cloud.google.com/vertex-ai/generative-ai/docs/partner-models/claude-prompt-caching

The command I'm running

VERTEXAI_PROJECT=redacted VERTEXAI_LOCATION=redacted aider --model vertex_ai/claude-3-5-sonnet-v2@20241022 --weak-model vertex_ai/gemini-1.5-flash-002 --dark-mode --cache-prompts --no-auto-commits

Version and model info

Aider v0.69.1
Main model: vertex_ai/claude-3-5-sonnet-v2@20241022 with diff edit format, infinite output
Weak model: vertex_ai/gemini-1.5-flash-002
Git repo: .git with 68 files
Repo-map: using 1024 tokens, files refresh