It should also be noted that no simple option exists within the Perplexity UI to export an entire session. Select-All/Copy/Paste captures only single or small groups. To test this: generate a session with a dozen or so pages (or open an extant session) and try to copy the content and paste to txt file. You'll see it doesn't work.
This does not apply to Claude, Gemini, Grok or GPT, all of which presently permit Select All
Edit: I also think it's worth discussing the actual advantages of using Perplexity over a dedicated model, eg Claude (my personal choice). Obviously, to offer so many models as Perplexity does, and for the same dollar amount, quality must be pruned somewhere. I have yet to purchase a subscription for a single model provider, but I am going to confidently guess that the session degradation, context preservation, memory and general coherence would be superior in a dedicated model. I'm not terribly informed on LLM details, but surmise some token austerity is involved in Perplexity's perplexing generosity. But I think it's safe enough to say, they must cut costs where they can, and the session export bug may rather be a feature of austerity, which I'm pretty sure is among more than a few more features disguised as bugs.
And yes, I'm mildly bitter. I've lost many hours of work due to being foolish enough to believe Export as PDF means export as pdf. A smart person doesn't believe what they read
It should also be noted that no simple option exists within the Perplexity UI to export an entire session. Select-All/Copy/Paste captures only single or small groups. To test this: generate a session with a dozen or so pages (or open an extant session) and try to copy the content and paste to txt file. You'll see it doesn't work.
This does not apply to Claude, Gemini, Grok or GPT, all of which presently permit Select All
Edit: I also think it's worth discussing the actual advantages of using Perplexity over a dedicated model, eg Claude (my personal choice). Obviously, to offer so many models as Perplexity does, and for the same dollar amount, quality must be pruned somewhere. I have yet to purchase a subscription for a single model provider, but I am going to confidently guess that the session degradation, context preservation, memory and general coherence would be superior in a dedicated model. I'm not terribly informed on LLM details, but surmise some token austerity is involved in Perplexity's perplexing generosity. But I think it's safe enough to say, they must cut costs where they can, and the session export bug may rather be a feature of austerity, which I'm pretty sure is among more than a few more features disguised as bugs.
And yes, I'm mildly bitter. I've lost many hours of work due to being foolish enough to believe Export as PDF means export as pdf. A smart person doesn't believe what they read