[Feature Request] Stream model output in real-time during generation

Bug Description
Feature request, I would like to see the model stream. It is nice to see the text arrive so I can read it while it is processing. Just watching the tokens count up is not very interesting. Things like Cline and Roo Code and Kilo Code are much more useful in the sense that I can read it as it arrives instead of having to wait for thinking or token output to finish and then press control-O. You will of course have to deal with terminal flicker and scrolling so I can scroll while the output happens. Maybe this is why you have not fixed the problem yet, but would very much like to see this as a feature.

Environment Info

• Platform: linux
• Terminal: vte-based
• Version: 2.0.27
• Feedback ID: 88db7009-288e-4967-aee6-39e5999e7574

Errors

[]