Added suport to anthropic prompt cache

[Claudio-code]

Thank you for taking time to contribute this pull request!
You might have already read the [contributor guide][1], but as a reminder, please make sure to:

issue #1403

I added support to user prompt use cache, added test only to api because Chatmodel/AnthropicChatOptions not can set header beta to works cache

• Sign the contributor license agreement
• Rebase your changes on the latest main branch and squash your commits
• Add/Update unit tests as needed
• Run a build and make sure all tests pass prior to submission

[markpollack]

thanks, can you explain the comment "Chatmodel/AnthropicChatOptions not can set header beta to works cache" please?

[Claudio-code]

Anthropic models added caching as a beta feature that requires a custom header to be used. I couldn't find a way to send a custom header using the options in the Chatmodel/AnthropicChatOptions classes.

That's why I added integration tests only in the AnthropicApiIT.java class.

It would be nice if in the future there was an option to send a custom header in ChatModel class to test beta features like langchain4j langchain4j/langchain4j#1591 (comment).

With this type of option it would be easier to add integration tests for beta features in the AnthropicChatModelIT.java class.

[markpollack]

Hi. Thanks for the clarification. I think it is a bit too much to get into M3 only because it modifies the AbstractMessage class and I'd like to think through the implications. I guess much like function calling, this is going to "table stakes" for leading model providers. I'll schedule it for M4 with the hopes we have a 5-ish week release cadence on our march to GA.

thanks for the nice PR.

[RRajdev]

@Claudio-code @markpollack @tzolov any chance this be merged any sooner than March?

[piratovi]

We are waiting for prompt cache feature too. It should help us to save tokens. Our app sends screenshots to Anthropic and we chose Spring AI framework to deal with LLMs

[Claudio-code]

I saw that the cache left beta, now I can add more complete tests to validate the creation of token caches and the reading of these caches.