Perplexity vs Copilot: Which One Should You Actually Use?

Perplexity and Microsoft Copilot both answer questions from the web with citations, and both run AI models from other companies. So Perplexity vs Copilot comes down to who gets to pick the model. On Perplexity, anyone paying $20 a month chooses from a table of models built by Perplexity and seven other companies, OpenAI, Anthropic and Google among them. On Copilot, Microsoft chooses. Its own pricing FAQ describes Copilot as bringing together "AI models selected by Microsoft within a single experience".

Most comparisons you will find frame this as an enterprise purchase, $30 a seat against Perplexity Enterprise, and list models that retired long ago. This one is for a person choosing for themselves: what each costs, where the models come from, which plan unlocks the research agent, and what you actually get on a Mac. Every price and feature comes from Perplexity's and Microsoft's own pages, checked on 26 September 2026.

The Key Takeaways

• Both cite their sources: Microsoft documents hyperlinked citations under Copilot's web answers. The real difference is that Perplexity searches by default, while Copilot searches when a question calls for it.
• Only Perplexity lets you pick the model: Pro at $20 a month opens a picker with models from OpenAI, Anthropic, Google, xAI and others. Copilot chooses for you, except in Excel on Microsoft 365 Premium.
• Copilot is cheaper to start: Microsoft 365 Personal is $9.99 a month and includes Office and 1 TB of storage. But Microsoft lists its Researcher agent as not available below Premium at $19.99.
• Agents cost extra on Perplexity: Pro includes no monthly credits for Computer, its agent, beyond a possible one-time bonus. Max at $200 includes 10,000 credits a month.
• The Mac apps are opposites: Copilot is in the Mac App Store and needs macOS 14. Perplexity's new Mac app is a direct download, not an App Store app, and on Pro or Max its agent can work in your local files and apps.

Perplexity vs Copilot: The Short Answer

Pick Perplexity if your questions end in research you have to check. You get sourced answers by default, models from several companies under one subscription, and a research mode with the $20 Pro plan. Pick Copilot if you already live in Word, Excel and Outlook, or if $9.99 for Office plus an assistant is better value to you than $20 for an assistant alone.

The deciding question is simple. Do you want an answer engine you steer yourself, or an assistant that turns up inside the apps you already pay for? The rest of this comparison is the evidence for each side.

What Each One Actually Is

The two get lumped together because both answer questions in a text box. They are built around different defaults, and the defaults decide most of what follows.

Perplexity is a search engine that answers

Perplexity describes itself as combining "web search with models built by Perplexity and by leading third-party providers". Search is not an option you switch on. It is the product, and answers come back with numbered sources you can click through. Around that sit a Research mode for long reports and a Learn mode for studying. Max adds Model Council, which puts one question to several models at once, and there is also the Comet browser and an agent called Computer.

Copilot is an assistant that lives in Microsoft's apps

Microsoft Copilot is the assistant Microsoft puts everywhere it can: a free app, a sidebar in Edge, and a button inside Word, Excel, PowerPoint and Outlook for Microsoft 365 subscribers. If you used Bing Chat, Microsoft says that experience "evolved into Microsoft Copilot". It writes, summarises, creates images, talks by voice and, when a question needs current information, searches the web. The name also covers a work product and GitHub's coding tool, which our guide to every version of Microsoft Copilot separates out. This comparison is about the consumer Copilot.

Who Picks the Model

Neither company relies only on its own model. Perplexity runs its in-house Sonar alongside other vendors' models. Microsoft's Transparency Note for Copilot says Copilot uses "language and multi-modal models from Microsoft, OpenAI, and other model developers". The difference is how much of that you control.

Perplexity hands you the picker

Perplexity publishes a dated table of the models in each plan. The snapshot of 4 September 2026 lists ten models from Perplexity and seven other companies: OpenAI, Anthropic, Google, xAI, Moonshot AI, Z.ai and NVIDIA. Pro gets eight of the ten. Max adds the two top models from OpenAI and Anthropic. On the free plan you get the automatic Best option and "a limited number of Pro Searches, but you cannot choose an advanced Search model manually."

Perplexity is also candid that a borrowed model is not the original product. Its FAQ answers "Are third-party models the same as using the provider's own app?" with a plain no. It explains that it integrates them with "its own search, citations, tools, prompts, safety systems, and usage limits". You get the vendor's model inside Perplexity's search, not the vendor's own chatbot.

Microsoft picks, with one exception

The consumer Copilot app has no model menu, and Microsoft's own pages describe that in two different ways. The pricing FAQ says "AI models selected by Microsoft". The Transparency Note for individuals, dated 18 August 2026, says "You can also choose between leading AI models for different kinds of work. Specific features vary by subscription."

In practice the choice lives in one place. Microsoft's Copilot in Excel FAQ says Excel supports Claude and GPT models. It adds that "using the model switcher requires a commercial Microsoft Copilot subscription or a Microsoft 365 Premium subscription." Even then, the choice lasts only for the session: when you close Excel, "Copilot reverts to the default model." We cover that switcher in detail in our Claude vs Copilot comparison. For everything else, and for everyone on Personal or Family, Microsoft decides.

Whichever you use, do not trust anyone's list of model names, including Perplexity's own. Its Pro help page, last modified on 3 September, still names older models than the table it links to. Its FAQ says "the list in your model selector will not stay fixed." The picker inside your account is the only list that is current.

Both Cite Sources, but Only One Always Searches

The most repeated claim on page one is that Perplexity cites its sources and Copilot does not, or does it badly. Microsoft's documentation says otherwise. The Transparency Note says that when Copilot's answers are grounded in the web, "the output contains hyperlinked citations listed below the text". Microsoft's support page on the free version lists "real-time answers based on web search" among its features.

What differs is when searching happens. Microsoft says that "for certain conversations where users seek information, Copilot is grounded in web search results". It is an assistant that searches when the question calls for it. Perplexity starts from the search, so even a casual question comes back as a sourced answer. If you want to check every claim, that default matters more than any model. If most of your questions are "rewrite this" or "draft that", you rarely need a search at all, and Copilot's habit of answering from the model is all you need.

Perplexity vs Copilot: Prices and Plans

Perplexity's prices come from its App Store listing and its Max help page. Microsoft's come from its Microsoft 365 plans page, which lists both monthly and yearly prices.

Two details change the arithmetic. On the Microsoft side, AI features "are only available to the subscription owner and cannot be shared with others", so the Family price does not buy Copilot for the family. On the Perplexity side, a Pro subscription covers searches and research but not much agent work, which the next sections come back to. The full breakdowns are in our Perplexity pricing guide, our Microsoft Copilot pricing breakdown and the explainer on what free Copilot gives you.

Long Research Reports: Research vs Researcher

Both companies now have a mode that goes away, reads many sources and comes back with a report. The price is almost the same. What differs is which plan you have to be on to get it.

Perplexity's Research mode produces "in-depth reports with more sources, charts, and advanced reasoning", and Perplexity picks the models for it automatically. Perplexity's plan comparison table gives free users one Research query a month and Pro users far more, so the report tool is really part of the $20 plan.

Microsoft retired Deep Research from the consumer Copilot app on 18 August 2026, and its report agent is now Researcher. The free Copilot app no longer has one at all. Microsoft's own AI credits and limits table marks Researcher as not available on Personal and Family, and as extensive use on Premium and Pro. It lists the same for Analyst, its data agent. So the $9.99 plan that makes Copilot look cheap does not include its report agent. The cheapest way to get it is $19.99 a month, a cent less than Perplexity Pro. At that price Microsoft adds the Office apps and Perplexity adds a choice of models.

For a tool-by-tool method rather than a plan comparison, our guide on how to do deep research with AI walks through the workflow whichever tool you use.

Agents and Browsers

Both companies are pushing past chat into assistants that do the work, and both have put that work into a browser.

Perplexity Computer runs on credits

Computer is Perplexity's agent for long, multi-step jobs: building a report, a dashboard or a spreadsheet from several sources. It is paid for in credits, and "100 credits equals $1". The catch is on the plan page: consumer Pro has "no monthly allocation". New subscribers may get a limited-time bonus of 4,000 credits, which expires after 30 days. Max comes with 10,000 credits a month. Ordinary searches never use credits, so this only matters if you plan to hand Perplexity real tasks.

Microsoft's equivalent is Cowork, which replaced Copilot Tasks. For personal accounts it is still in preview, and Microsoft calls it "a Premium or Pro feature", so Personal and Family subscribers do not get it.

Comet vs Copilot in Edge

Perplexity's Comet browser has been free for everyone since 2 October 2025. Perplexity's launch post says every new tab has a Comet Assistant "to answer any question or take any action on your behalf." Its current plan table, though, gives free users no browser-agent queries. Those start with Pro.

Microsoft went the other way and folded Copilot into the browser it already had. On 13 May 2026 it retired the separate Copilot Mode and built its features into Edge itself, on Windows and Mac. That includes Copilot reading across your open tabs, which Microsoft says is now available to everyone. The agent part, Browse with Copilot, is narrower: Microsoft 365 Premium subscribers in the U.S. only, with usage limits. So both browsers are free and neither agent is: Comet's starts at Pro for $20, and Edge's at Premium for $19.99, in the U.S. only. Our roundup of the best AI browsers sets Comet against Chrome, ChatGPT Atlas and the rest.

Perplexity and Copilot on a Mac

This is where the two differ most, and where most comparisons say nothing.

Microsoft Copilot is a normal Mac App Store app from Microsoft Corporation. It needs macOS 14 Sonoma or later.

Perplexity replaced its Mac app on 7 May 2026 with a new one built around Personal Computer, a version of its agent that runs on your Mac. According to Perplexity's announcement, it can work "with local files and native Mac apps" and hand web tasks to Comet. Perplexity lists a Command + Command shortcut to open it. Anyone can use the app for everyday questions, but Personal Computer itself needs Pro or Max, and its work uses your plan's credits. Setup asks for Accessibility, Screen & System Audio Recording and Full Disk Access permissions, and Perplexity says it works best on an always-on Mac mini. Two practical warnings. Perplexity's download page says it needs macOS 15 Sequoia or later, while its help centre still says macOS 14 Sonoma, so check before you count on an older Mac. And it is not in the Mac App Store: you download it from perplexity.ai. Searching the Mac App Store for Perplexity turns up a similarly named app from a different developer, so do not install the first result.

So on a Mac, Copilot is the easier install and the lighter app. Perplexity is the more capable one if you want an agent that can act on your own files, provided you pay for Pro or Max and your Mac meets its requirement. We cover the Perplexity side in more depth on our page about Perplexity on a Mac.

Which One Should You Pick

Pick Perplexity if you research for a living or a hobby and want every answer sourced by default. It also suits you if you want to test the same question on models from different companies without paying each of them, or if you want Comet's browser agent, which comes with Pro.

Pick Copilot if you already pay for Microsoft 365, or would buy it anyway for Word, Excel and storage. At $9.99, Personal is the cheapest paid plan in this comparison, and Copilot turns up inside the documents you are already working in.

Pick Microsoft 365 Premium over Perplexity Pro only if you want Researcher, the Excel model switcher and Office in one bill, since at $19.99 against $20 the price is a wash.

The same comparison against OpenAI is in Perplexity vs ChatGPT and Microsoft Copilot vs ChatGPT. Against Google, see Perplexity vs Gemini and Copilot vs Gemini.

Or use Perplexity without choosing a side

If the appeal of Perplexity is choosing the model, you can go one step further on a Mac. Fello AI puts Perplexity next to GPT, Claude, Gemini, Grok and DeepSeek in one native app, with web search, PDF chat and Word, Excel and PowerPoint files. There is a free tier, and it runs on Mac, iPhone and iPad.

The Verdict

Perplexity wins on research and on choice. It searches by default, cites its sources, lets a $20 subscriber choose among eight models from several companies, and gives away its browser, with the browser agent included in Pro. Its weak spot is agent work, which Pro barely covers without buying extra credits.

Copilot wins on price and placement. For $9.99 you get an assistant inside the Office apps plus the apps themselves. Copilot searches and cites when it needs to, but Microsoft chooses the model, and its research and browsing agents start at Premium.

If you pay for Microsoft 365 already, start with the Copilot you have and add Perplexity only when you catch yourself checking its sources by hand. If you do not, Perplexity Pro is the better $20 for anyone whose questions usually start with where the evidence is.

Frequently Asked Questions