I'd love to say "just use venice" but honestly, it's not always great on mobile. The PWA is very laggy and buggy.
I'm still using perplexity for some things, but I'm currently switching to litellm connected to all the APIs. That way I save a lot of money and I'm still able to use the best models, just using their API.
Venice, especially the qwen3 235b (venice large) is really nice. Even if the PWA sucks, you can use it through your own apps like openwebui, as long as you're ready to get some venice tokens staked for a good amount of VCU.