When will any publicly available consumer AI assistant (ChatGPT, Claude, Gemini, etc) be able to look up a public domain book of reasonable length (~100k words) and read it (all of it) aloud to me in one shot.

A few clarifications:

• Multiple options can resolve yes

• This applies to assistants where you don’t self-host models on your own GPUs

• It should be able to do this after a single request from the user, without requiring repeated prompts (feel free to prompt engineer that starting prompt as much as you'd like). In other words, the audio it plays should be like what would happen if you pressed play at the start of an audiobook

• The product should be able to do this relatively consistently, >~ 50% of attempts

• The model has to be able to source the text from the internet. You can provide a link to a website with the text, but can't give the model a pdf (or other file) containing it in your prompt

• An AI assistant that, after one user request, fetches the public-domain text in chunks internally, streams a TTS tool continuously, and plays the whole book without further prompting would count

Here's conversations with ChatGPT, Claude, and Gemini I had today/at market creation (Apr 24, 2026) where they refused and/or failed (props to Claude for the effort though). All of them happened in voice mode:

ChatGPT:

Gemini:

Claude:

^ It got the first few paragraphs in its last try, but then started making things up