Perplexity

AI search engine that gives you sourced answers instead of a list of links. Still one of the best dedicated research tools, but the research built into Claude, ChatGPT and Gemini now covers what most people need.

Free / $20/month Pro / $200/month Max

My take

Perplexity used to be indispensable for me. Then the research built into the big assistants got good enough that most people don't need a separate research tool anymore, and in my AI circles hardly anyone talks about Perplexity now. It didn't help that they cut the Pro limits and tried to become too many things, including an app builder that costs far more than the dedicated ones. I have no big research projects right now, so my subscription is on hold. But I'm not writing them off: if you research for a living or need sources you can really rely on, a dedicated research tool is still worth paying for. Four stars, still solid.

Visit Perplexity Back to tools

At a Glance

Pros

+ Every answer comes with cited sources you can verify
+ Deep Research is fast and stays close to its sources
+ Recurring research tasks (e.g., daily media monitoring)
+ Excellent for learning new topics quickly
+ Clean, focused interface

Cons

− Pro limits were cut more than once in 2026; heavy Deep Research use now pushes you toward the $200/month Max plan
− Tries to be too many things at once, including an app builder that costs far more than dedicated builders and does less
− The research modes in Claude, ChatGPT and Gemini are now good enough for most people, so it's often a second subscription you don't need
− Sometimes over-relies on a few sources

Best for: People who research for a living and need sources they can rely on: market and competitor analysis, fact-checking, regulatory questions

Full Review

Perplexity is what Google should have become. Instead of giving you ten blue links, it reads across multiple sources and gives you a direct, sourced answer. For a long time it was my first stop for anything that needed sources. That has changed, and not only because of Perplexity.

Why I downgraded it to four stars

The competition caught up. Claude, ChatGPT and Gemini all have research modes built in now, and they’re good. For most people and most questions, the assistant you already pay for will do the research, and a second subscription just for research is hard to justify. You can see it in the conversation, too: in the AI circles I follow, Perplexity has mostly disappeared from the discussion.

Perplexity made itself harder to recommend. The Pro plan’s limits were cut more than once this year, and if you lean on Deep Research, you run into them quickly. The upgrade on offer is the $200/month Max plan. At the same time, Perplexity tried to become a browser, an agent and an app builder. I did build a small weekend app with its agent. It worked, but the free credits ran out halfway and I had to buy more. For that money, the dedicated app builders give you more, and they’re more capable.

So my subscription is on hold. I have no big research projects right now. When I do, I’ll probably be back.

Who should still pay for it

If you research for a living, a dedicated research tool is still worth having. Market and competitor analysis, regulatory questions, anything that goes into a document other people will rely on: that’s where Perplexity’s habit of tying every claim to a source you can click pays off. When I researched the Swiss labor regulations for temporary staffing, I used Perplexity and prioritized official government sources, because the result was going into a knowledge base people would act on.

If you have a research question once a week, use the research mode in the assistant you already have.

How I used it

Quick research. When I needed to understand a topic fast, Perplexity gave me a well-structured overview with sources in seconds. A new technology, a market trend, a regulatory question.

Deep Research for strategic work. Give it a complex question and it spends several minutes running searches, reading dozens of sources, and writing up a full report. I used it for competitive analysis, understanding potential customers, and market sizing, usually as the first step of a pipeline that hands the results to Claude.

Recurring tasks. You can set up research that runs on a schedule. I had daily media monitoring running that tracked industry news, competitor moves, and market developments, delivered automatically.

Practical use cases

Market research: “Who are the main players in [your market], and how do they differentiate?”
Learning: “Explain how transformer architectures work, starting from the basics”
Competitive intelligence: “What has [competitor] announced in the last 30 days?”
Technical evaluation: “Compare the latest approaches to AI-powered matching algorithms”

Where Perplexity is heading

Most of the new work goes into Perplexity Computer, their agent. Automations arrived on 5 October 2026: they run on a schedule or when something happens in a connected app. In September it also got a hybrid mode on Mac that mixes cloud and local models. Its memory feature, Brain, belongs to Computer too, as a Max-plan feature in research preview. The search side keeps adding the newest models: GPT-6 Astra in September, then GPT-6.1 Sol and Claude Opus 5.5 on 5 October. If you only want research, the $20 plan ($17/month billed annually) is the one to look at (Max costs $200/month or $2,000/year). Check the limits in your account before you commit to a year: they have moved before.