Perplexity AI

Pros and cons

How Perplexity differs from ChatGPT and Claude

Perplexity is built for research, not conversation. When you ask a question, Perplexity searches the web, reads the relevant results, and writes a synthesised answer with numbered citations alongside every claim. You can click any citation to read the original source.

[ChatGPT](/tools/chatgpt) and [Claude](/tools/claude) are general-purpose assistants — they can write, code, reason, and hold long conversations. Perplexity does one thing: it finds current information and shows you where it came from. For that specific task, it is faster and more transparent than the general-purpose tools.

The free plan on [Claude](/tools/claude) does not include web search. The free plan on [ChatGPT](/tools/chatgpt) includes limited web search. Perplexity's free plan has real-time web access by default. If you regularly need current information — news, recent product releases, current pricing, recent research — and you want that information with citations, Perplexity is the direct tool for it. For anything beyond research — writing assistance, coding, complex reasoning — it is not the right choice.

How the citation model works

Every answer Perplexity generates includes numbered superscripts. Each number links to the source Perplexity used for that claim. At the bottom of the answer, all sources are listed with titles and URLs. You can open any of them with one click.

This citation model reduces the risk of the most common AI problem: confident-sounding hallucination. When a claim is sourced, you can check whether the source actually says what Perplexity claims it says. When no source appears for a claim, that is a signal to be sceptical.

The limitation is that Perplexity's sources are not always the best available. It tends to cite whichever pages rank highest on Google for the query, which often means news sites, Wikipedia, or content marketing articles rather than primary research or authoritative data. For high-stakes fact-checking — verifying a medical claim, checking a legal precedent, confirming a technical specification — treat Perplexity as a starting point that tells you where to look, not as the final answer. For general research and current events, the citation model is substantially more useful than AI tools that produce answers with no attribution at all.

What Perplexity Pro adds

Perplexity Pro ($20/month) adds four meaningful features: model choice, Deep Research, file uploads, and image generation.

Model choice lets you select GPT-4o or Claude for answers instead of Perplexity's default model. This is useful when you want the citation-based search format combined with a more capable underlying model for complex synthesis. Deep Research is the most practically useful Pro feature — it runs a structured multi-step research process, reads dozens of sources, and produces a long-form report with citations throughout. For research projects that would otherwise take hours of manual reading, Deep Research compresses the process significantly.

File uploads let you ask questions about PDFs and documents, with citations pointing to specific sections of the file. Image generation is included but is not Perplexity's focus — the output quality is below dedicated image tools. If you need image generation, [ChatGPT](/tools/chatgpt)'s DALL-E integration is more capable. At $20/month, Perplexity Pro costs the same as [Claude](/tools/claude) Pro and [ChatGPT](/tools/chatgpt) Plus. The choice between them is a function of your primary use: research and current information goes to Perplexity; writing, coding, and general tasks go to Claude or ChatGPT.

Spaces — shared research workspaces

Perplexity Spaces are shared research environments where teams can collect sources, share AI-generated summaries, and ask questions against a shared knowledge base. You add websites, files, or Perplexity-generated research to a Space, and all members can query that material directly.

For small research teams — journalists, analysts, competitive intelligence functions — Spaces provide a lightweight way to pool research without a full-blown knowledge management system. You collect the relevant sources in one place, and the AI can answer questions that draw on all of them together.

Spaces are available to Pro users. The practical limitation is scale: Spaces work well for a defined research project with a specific scope. They are not designed as long-term knowledge repositories for large teams. For that, tools like Notion with AI search or a proper knowledge management system are better suited. Perplexity Spaces solve a narrower problem — collaborative research on a current, bounded topic — and they solve it cleanly.

Who should not use Perplexity

Perplexity is a research tool. It is not a writing assistant, a coding tool, or a general-purpose AI. If your primary need is generating long-form written content — articles, reports, marketing copy, documentation — [Claude](/tools/claude) is a substantially better tool. Perplexity generates summaries and research notes; it does not produce polished prose.

If you write code professionally, use [Cursor](/tools/cursor) or [GitHub Copilot](/tools/github-copilot). Perplexity can answer technical questions with citations, which is occasionally useful for debugging or finding documentation, but it is not a substitute for an AI tool designed for coding.

If you need image generation, Perplexity's image output (Pro only) is not competitive with dedicated tools. Pay $20/month for [ChatGPT](/tools/chatgpt) Plus if image generation is part of your regular workflow — DALL-E 3 is integrated and far more capable.

For users whose primary need is to ask questions and get good answers — not research reports with citations, just answers — [Claude](/tools/claude)'s free plan or [ChatGPT](/tools/chatgpt)'s free plan covers most of that without a paid subscription. Perplexity's free plan is worth using for anyone who frequently needs current information with sources. The Pro plan is only justified if you regularly need Deep Research, model choice, or Spaces — casual research is well served by the free tier.