[VHRanger]

It uses only google as a backend, and the answer engine is based on the ancient llama 3.3 70B.

It's well behind Kagi (which uses a bunch of sources for results and new models for quick answer or assistant). Perplexity is even well behind other competitors like chatgpt

[unixhero]

Ok ok, I'm signing up for Kagi now

[QuantumGood]

Perplexity Pro allows you to use GPT-4 Omni, Claude 4 Sonnet, Gemini 2.5 Pro, Sonar, DeepSeek R1, and others. Can you clarify what you mean by based on the ancient llama 3.3 70B? And they pull results from multiple search engines, and their own index, not just Google.

Preference doesn't have to be based on anything, but providing misinformation that is easily verified is not that common on HN. And Perplexity Pro is different than vanilla Perplexity.

[VHRanger]

I was talking about their sonar API and what you get when you go to perplexity.ai and search by default -- it's llama 3.3 70B hosted with cerebras. Which is fast, but way way behind the times -- it's more likely to hallucinate than models from basically anyone from the competition, including quick answer.

You're right though it seems that they don't just use bing + their own crawler, but who else they use seems to be community speculation.