feat: Add conditional dropdowns for Verbosity and Reasoning Effort if model is supported

[HarryVu176]

Some OpenAI models support Verbosity / Reasoning Effort parameters. I think it will be handy to be able to change those settings on the toolbar for quick (especially reasoning effort) instead of having to go to model dropdown -> model settings.

[vercel]

@HarryVu176 is attempting to deploy a commit to the Token Fabrics Pro Team on Vercel.

A member of the Team first needs to authorize it.

[HarryVu176]

I have attached the image here

[enricoros]

Hi @HarryVu176 - this is a great change and well done. I can se why it's convenient to have params there.
Please take a look at https://app.big-agi.com, which I've just dark launched, and has parameters setting in the convenient side-pane. May that be useful to you, or do you prefer in the dropdowns?

[HarryVu176]

Hi @HarryVu176 - this is a great change and well done. I can se why it's convenient to have params there. Please take a look at https://app.big-agi.com, which I've just dark launched, and has parameters setting in the convenient side-pane. May that be useful to you, or do you prefer in the dropdowns?

Thanks a lot for checking the PR! I'm no UI expert, I tried agi.big-agi.com and it looks polished. On my vertical monitor/small screen, I rarely toggle the right panel, I usually leave it closed, so having params in the dropdowns works well. I think it also depends on how often people change verbosity/reasoning, I tend to switch between depending on my question/context to save tokens.
That's what I thought, but I will follow your design guidelines and decisions!

[DoS007]

Easy switching for reasoning or websearch is relevant (didn't test new github repo changes).

Include reasoning and websearch in message author name: I also think, that knowing which answer was given with how much thought or sources (reasoning or websearches, but especially reasoning as you cannot see it as directly) is relevant.

Btw. when using Beam and take all messages without combination it would be nice to see each answer better or see which models were used instead of "unk-model". And i believe when the first message is beam the chat name doesn't get genereted automatically.