It's preatty easy to hit the token limit per minute. A way to save some tokens might be reducing the context. here some ideas: - images link - available_filters