I asked the model about its capabilities, and it turns out it indeed can do Web searches; if it's not hallucinating, the backend server indeed decrypts the outp

I asked the model about its capabilities, and it turns out it indeed can do Web searches; if it's not hallucinating, the backend server indeed decrypts the output of the LLM; only the user prompt is E2EEed against the server Edit: I'm a little weary to find there is convenient import but not export functionalities. I manually copied the conversation into a markdown file <https://gist.github.com/Gravifer/1051580562150ce7751146be0c98ded9 https://gist.github.com/Gravifer/1051580562150ce7751146be0c9...>