I use Sonnet 4.5, mainly because that’s what I use independently and am happy with the results. I don’t use anything from OpenAI because they are sell outs with no sense.
it seems the list from perplexity is focused on live audio processing, which I don’t think is relevant here.
I really don’t think it matters that much … whatever template or system prompt will have a bigger influence.
Got my pocket the other day, best models available for me at the time were Claude 4 Sonnet and GPT-4o, but just now GPT 5 has appeared as an option. I’m rolling with GPT 5 at the moment, though I personally use Gemini most of the time so wouldn’t mind that as an option. Well see how they go during my meetings.
Me personally, I would rather have an option to download an LLM or use Apples AFM on my iPhone. That way I control my data and it doesn’t leave my device unless I explicitly say so.
You mean in general within everyone’s overall workflow? Or in Pocket? I’m guessing in general, since the Pocket options are,
Gemini 3.1 Pro Preview (Google) - Pro
Claude Opus 4.8 (Anthropic) - Pro
Gemini 3 Flash Preview (Google) - Core
GPT-5.6 (OpenAI) - Core
Claude Sonnet 5 (Anthropic) - Core
In general, I use GPT for structured reasoning since it seems better at preserving overarching context, and Gemini for large document-based (pdf) research.