This is a big one for me and currently holding me back from using claude with typing mind. Its too expensive without prompt caching, and using anthropics API directly is much worse than open router.

Support this suggestion! Prompt caching is indeed a very practical feature, especially for large models like Claude. This not only can:

Significantly improve response speed

As a frequent user of TypingMind, I also look forward to this feature being implemented as soon as possible. This will make the integration of TypingMind with OpenRouter more complete.

I believe e have this already? I can activate it in TM Custom at least