“Perplexity augments fast and cheap models like GPT 5.6 Luna to improve accuracy and overcome hallucinations. Now compatible with any open harness: via CLI, SDK, or MCP.”
“@ArtificialAnlys @perplexity_ai The interesting signal is the payload efficiency: Perplexity’s lead isn’t just retrieval quality, but how little context the model needs to reach the answer. That makes search quality and inference economics increasingly inseparable.”
“I'm not saying Perplexity is a Ferrari, but he seems unfamiliar with the tool he's using and is comparing it to a similar AI system. Maybe I'm mistaken; I don't know who he is.”
“This is the difference between using Perplexity Search and Perplexity Computer. It’s common for people to not realize they are using the Computer part of the app which uses credit you pay for outside the subscription.”
“perplexity is different though, they seem to scrape everything and youtube transcripts just happen to be dense with step-by-step instructions. the "answer-shaped content" thing you said makes sense, a video tutorial is basically someone talking through a problem for 10 minutes, that's way more signal than a random blog post trying to rank for keywords”