Curious - Does BAML support hooking into gemini cache https://ai.google.dev/gemini-api/docs/caching?lang=rest --- If not, are there any suggestions for caching which may get me part of the way? I read somewhere that it will automatically cache your prompts simply by ensuring the static text remains at the top of the prompt & dynamic parts sit below it which results in cached tokens for the static part 🤔 Haven't verified this myself

you get by default: Implicit caching (automatically enabled on Gemini 2.5 models, no cost saving guarantee) you can also add anything you want to the api call. So just pass the cache content yourself