Hi,
Small suggestion for the reflective models, especially during deep researches: it would be quite nice if the reflection text (the one that appears before the final answer) could be read aloud using the chosen voice as it appears.
Since TTS is already there, and the reflection shows up progressively, it seems like everything’s in place, minus a few tweaks to auto-play the voice for each new chunk of reflection, without needing to click “play” every time.
It would add a nice layer of immersion. Like hearing a collaborator think out loud.
Nothing critical, just a neat little option to get even closer to that human-like assistant feel.
Thanks!