One-line positioning
It rebuilt "search" from scratch: answers come with citations and can act across your open tabs
Perplexity is the product that dropped a large model into the search box. It doesn't chit-chat — its core job is: you ask a question, it goes online, pulls a pile of sources, and returns a cited answer so you skip clicking ten links yourself. In 2026 it doubled down on the Comet browser, trying to move the assistant out of the chat window and into the page you're already reading.
Getting Started (based on community-tested notes, not first-person)
Note: compiled from official docs and public reviews, rewritten and condensed by editors — not a claim that we tested each point ourselves.
- Pick an entry point: for pure research use the web app or the App's Spaces (persistent research rooms where you upload files and set instructions); to let the assistant read the page you're viewing, install the Comet browser.
- Use the right mode: for cited synthesis turn on Deep Research; for price comparison, form-filling, and cross-site checks use Comet's agent; for quick factual lookups use normal search.
- Watch the tiers: the Comet browser itself has been free on all platforms since Mar 18, 2026 — core agentic search, voice, and Deep Research are all free; you only step up to Max at $200/mo for Background Assistants and the multi-model Model Council.
Real Pain (community consensus)
Note: recurring pain points reported across the community, condensed by editors, not individually verified by us.
- Citations ≠ correct: a Columbia Journalism Review audit found about a 37% error rate in standard answers — a source existing doesn't mean the claim holds, so verify the key conclusions by clicking through.
- Automated shopping carries legal risk: in Jan 2026 Amazon sued Comet over auto-checkout; the feature is live for now but sits under a legal cloud — don't let it pay for you.
- Weak on local / transactional: restaurants, maps, price comparison — it lags Google there, and text-only shopping research trails commerce sites.
- Max is expensive: the "let it run free" autonomous agents need $200/mo; light users are fine on the free tier.
Hard Comparison (one-line verdicts)
- ChatGPT: wider on long-form writing and agent ecosystem, but Perplexity's "answers always cited, sources checkable" is the root difference.
- Gemini: strong on Google ecosystem and huge context, but Perplexity's cross-source synthesis and citation experience are more research-focused.
- Domestic assistants (Tongyi / Doubao): free and localized in Chinese, but Perplexity leads on global source coverage and citation rigor.
Who It's For / Not For
- For: research synthesis, citation-backed writing, cross-tab comparison, analysts / students / writers.
- Not for: heavy local-life search, those wanting it to auto-checkout, or strict scenarios with zero tolerance for citation error.
Editor's Take (our site's view)
Note: our judgment from a design-practitioner / AI-video angle — the angle that sets us apart from repost-only review sites.
Perplexity is our "find sources, find provenance" blade — pulling references before a draft, competitive comparison, and grounding client pitches. It beats blindly asking ChatGPT because every line links back so we can verify. But we never treat it as settled truth: a citation existing doesn't mean the claim is right, so any important number gets a click to the original link. We installed Comet; the side panel summarizing the page you're reading genuinely saves steps, but we never switch on the "hands-off agent" that checks out for you — Amazon's lawsuit is the warning. Our advice to small teams: use it as a cited research assistant, double-check the important claims, and don't treat it as a source of truth.
Ratings
Cited research: ★★★★★
Free availability: ★★★★★
Answer accuracy: ★★★☆☆
Ease of use: ★★★★☆