### What Comet is and what it is not Comet is Perplexity's AI browser, with a built-in assistant that answers questions about the page you're on and an agent that carries out multi-step tasks across tabs for you. It's an interface to Perplexity's service, not a separate product with its own meter. In the picture below, the three layers sit in order: - The browser at the front with no price attached - The Perplexity plan (Free, Pro or Max) behind it - An agent-task usage meter that the plan feeds  Read it right to left when you budget, since the plan sets the meter and the browser only displays it. That ordering also separates Comet from tools built to run automations on a schedule. Comet works while you sit in front of it. What follows prices the plans and the usage behind them. ## Every Comet plan with price, inclusions and limits ### The plan table Because Comet the browser is free to install, every paid line below is a Perplexity plan, and what separates them is how much assistant and agent work they allow. | Plan | Monthly price | Annual price | Inclusions | Deciding usage limit | |---|---|---|---|---| | Comet free | not published | not published | The browser with its built-in assistant | Capped assistant and agent use; cap not published | | Comet Plus add-on | not published | not published | not published | not published | | Enterprise Max | not published | not published | not published | not published |  Treat a blank in this table as a question for Perplexity's sales or billing contact, not as a zero. A plan whose price and allowance are both unpublished can't be budgeted from the page alone. ### Which limit decides your tier **It is not the browser that decides** your tier. The number of assistant and agent tasks you run each month does. Because Perplexity publishes few of these allowances, the practical test is your own usage: count the agent tasks your team runs in a normal week and note how many stall at the free cap. If the free tier covers you, nothing else is worth buying. If you hit the cap occasionally, Pro is the first step up. Max matters only when agent runs are routine rather than occasional. The Enterprise tiers belong to teams that need shared administration across seats. What follows turns these tiers into a monthly bill for one small team's workload. ## What the plans cost at realistic usage $0 on Free, **$60 on Pro, or $600 on Max**: that is the monthly bill for a three-person marketing team. The browser adds nothing to any of them, so the team's monthly cost is determined by plan choice alone.  The spread comes entirely from the plan and the agent work behind it. ### Example workload for Comet pricing estimates Every day, each person runs competitor research: a handful of assistant questions about pricing pages, launches and positioning. Once a week, an agent runs monitoring across the same competitors and reports what changed. The research is frequent but light, while the monitoring is rarer and heavier. That split decides which tier fits, because assistant questions and agent runs draw on usage differently. ### Monthly bill by plan Side by side, the chart shows the same team's monthly bill at each tier. Free costs nothing, Pro across three seats comes to $60, and Max across three seats comes to $600, ten times Pro, making tier selection a question of budget rather than capability.  Take from it that the jump from Pro to Max is the expensive decision. Moving up a tier costs far more than adding a seat does. That gap pays off only if the weekly monitoring runs actually hit Pro's ceiling. I have no confirmed limit to quote, so test it against your own runs before paying for Max. ### Cost per seat and per agent task $20 per seat on Pro and $200 per seat on Max is what the chart works out to. A seat that never exhausts Pro is wasted money at Max. I can't give a trustworthy cost per agent task, because it depends on how many runs a seat completes and on limits I haven't confirmed. You can calculate it for your team by dividing the seat price by the runs you actually complete in a month.  When daily research and weekly monitoring compete for the same allowance, Free runs out. Start there, note the week a task is refused, and upgrade only the seats that hit the wall. ## Factors affecting your Comet bill Expect the bill to move on three things: the usage ceilings that force an upgrade, whether you pay monthly or annually, and which seats sit on the pricier Enterprise tiers. ### Free tier and rate limits Budget on today's cap and you'll be wrong later: the free tier and the paid tiers all cap advanced searches, agent runs and premium model use, and [Perplexity](https://www.perplexity.ai/help-center/en/articles/13838041-how-credits-work-on-perplexity) adjusts those caps over time.  Your own log of refused tasks, one per week per person, tells you which seat has actually hit the wall and which is merely nervous. Pay annually and every seat price drops, according to [Metronome](https://metronome.com/pricing-index/perplexityai). The gap per seat per month: - Pro falls from $20 to $16.67, about $40 a year saved per seat. - Max falls from $200 to $166.67, about $400 a year per seat. - Enterprise Pro falls from $40 to $33.33, about $80 a year per seat. - Enterprise Max falls from $325 to $270.83, about $650 a year per seat. On Max seats the discount is worth most; on Pro, least. Commit annually only for seats that have already proven their usage, and keep new or uncertain seats monthly. Moving a whole team up for one person's needs doubles the line for everyone else, since the Enterprise Pro seat costs twice the Pro seat at monthly rates. Mix tiers by seat rather than standardizing. Treat data handling and admin controls as the reason to pay for Enterprise, not usage volume. If nobody on the team needs those controls, the extra $20 per seat per month buys nothing you'll use. If someone does, price that seat separately and leave the rest on Pro. ## Automate repeat Comet tasks with Activepieces If the job is one you keep handing to a browser agent every week, it probably belongs in a workflow instead of an agent run. ### Comet tasks better run as workflows ### Why repeat tasks fail in Comet's agent Most teams find this out by accident. Someone asks the agent to check competitor pricing pages on Monday, then again the next Monday, and soon it's a ritual nobody owns. Each run spends agent usage on the same steps, and each run can wander differently: a changed layout, a cookie banner, a page read slightly differently. Handing a task you repeat on a fixed rhythm with a fixed shape to the most flexible and most metered tool you have is a poor use of it. Run one test: **if you can write the steps down** in advance, the agent is paying for judgment the task never needs. ### Benefits of workflows over Comet agent runs Because **a workflow runs the same path every time**, the bill is predictable and a failure shows up at a named step instead of somewhere inside a long agent session. It also lets you grant each step only the access it needs, which is the least-privilege problem solved by design rather than by a policy line. The browser agent sits beside it for one-off tasks. Take from it that the recurring path and the exploratory path are different jobs, and each gets its own tool. Once the recurring work is out of the agent, what remains is the work that genuinely needs one. ### Running automated workflows on schedule Activepieces is an automation platform that builds this kind of flow: a trigger, a series of steps, an AI step in the middle, and integrations at the end. You set the schedule once, and it runs without anyone opening a browser. Its published pricing page lists 1 credit per flow run, however many steps the run has, whereas Zapier's per-task billing on its own pricing page counts each step separately. So the weekly fetch, summarise and post flow costs one credit, the same as a two-step version of it. You don't need it. That holds when the task is truly one-off, when the steps change each time, or when a single scheduled prompt in a tool you already pay for does the job. ## Comet compared with the closest AI browser alternatives Every rival here follows the same model as Comet, a free browser in front of a paid assistant plan. So the cheapest choice is usually the plan you already hold, and the seat you priced separately is the only one that should change that. ### Comparison table Here is what each browser costs to enter, which plan unlocks its agent, and where the limits sit. I left out prices because they move often, and the useful point is that each limit comes from the plan behind the browser, not the browser. It was read on 2026-11-10. | Browser | Entry price | Paid tier | Agent mode | Usage limits | Platforms | Check date | |---|---|---|---|---|---|---| | Comet | Free browser | Perplexity Pro or Max | Comet Assistant runs multi-step tasks | Set by the Perplexity plan | Desktop and mobile | 2026-11-10 | | ChatGPT Atlas | Free app | ChatGPT paid plans | Agent mode on paid plans | Set by the ChatGPT plan | macOS | 2026-11-10 | | Dia | Free | Dia's own paid tier | Assistant-led, lighter agent features | Set by the Dia plan | macOS | 2026-11-10 | | Gemini in Chrome | Free with Chrome | Google AI Pro or Ultra | Agentic features on paid Google plans | Set by the Google AI plan | Chrome desktop | 2026-11-10 | | Claude in Chrome | Included with a paid Claude plan | Claude paid plans | Browser agent in the extension | Set by the Claude plan | Chrome desktop extension | 2026-11-10 | Since every row's limits come from a subscription, compare plans, not browsers. When you already pay for its plan and your team is already using it unsanctioned in a personal tab, a rival is cheaper. Someone on ChatGPT, Google AI or Claude gets an agent for no extra line item. Adding a Perplexity seat for Comet then buys a second subscription for work the first one can do. Comet is worth it when research and search are the main workload and Perplexity is already the team's assistant. There the browser adds nothing to the bill, and the agent runs against the plan you already sized. If that describes your team, stay on Comet. If not, use the rival you already pay for. ## Choose your Comet tier **Start on the Free plan, count the agent tasks your team actually runs for a week, and pay for a higher Perplexity plan only where that count crosses a limit.** The browser costs nothing either way, so the log is the evidence. Most teams already have someone running Comet on a personal account and expensing nothing. That's useful, because it means you can measure real use instead of guessing at it. Through the first three steps you collect the evidence, and the last of them tells you whether the table's limits fit your workload or not. 1. Put everyone who'll touch Comet on the Free plan, using work accounts rather than personal ones, so usage lands in one place. 2. Keep a shared log for a week. For each agent task, record who ran it, what it was, whether it finished, and whether it hit a limit or got throttled. 3. Compare the log against the limits in the plan table earlier in this article. Note which limit you reach first, whether that's assistant queries, agent tasks, or something else. Hit a wall on day two and you learn more than the person who never notices one does. The wall's type is what decides the tier. Once you've moved off Free, the last three steps keep the spend from drifting. 4. Upgrade a single seat, the heaviest user in your log, and leave everyone else where they're. If the upgrade removes the friction, you've proof. If it doesn't, you've saved the cost of upgrading everyone. 5. Set a recurring reminder to recheck the Perplexity pricing page. Plans and limits change, and your table will go stale before your invoice does. 6. Write cancel criteria before you pay. For example: the upgraded seat ran fewer agent tasks than the Free limit would've allowed over a full month, or the tasks it ran were the one-off kind you decided not to automate. Write the criteria down and a later cancellation becomes a routine check rather than an argument. ## Frequently asked questions Once the criteria are written down, the same few questions'll come up again, and answering them in the document saves the argument later. ### Is Perplexity Comet really free? ### What the free browser does and does not include The browser is free to download and use. What you pay for is the Perplexity plan behind it, because assistant and agent tasks draw on that account's usage. A free account gets you in the door, but how far agent work goes before it hits a limit depends on the plan, so check Perplexity's current plan page before you budget. Not if you're asking whether you need Perplexity Pro to use Comet's agent. A free account can use the assistant within its limits. Pro matters when you run agent tasks often enough that those limits become the bottleneck. Count your weekly tasks first. If you rarely hit the ceiling, the upgrade buys you nothing. ### Education offers, work use and regional pricing Perplexity has run education offers on its plans, and they apply to the plan, not the browser. Eligibility rules and terms change, so confirm on Perplexity's own education page instead of relying on a third-party summary. Treat any discount as a promotion, not a permanent price. You can use Comet at work on a team plan, through Perplexity's business offerings, not through anything specific to the browser. The decision is the same tradeoff as for individuals: how many agent tasks the team runs. The extra consideration is what the agent can reach. Give it a work profile with only the accounts it needs, and decide that scope before anyone signs in. Where you are changes the price. Currency, taxes and plan availability vary by region, so the number on your checkout screen is the one that counts. Compare tiers using the price your account shows, not a figure from an article written for another market. ## Related reading - [Perplexity API Pricing in 2026: Models & Costs](https://www.activepieces.com/blog/perplexity-api-pricing-in-2026-models-costs) - [Clerk Pricing 2026: Free, Pro and Business Plans](https://www.activepieces.com/blog/clerk-pricing-2026-free-pro-and-business-plans) - [Genspark Pricing 2026: Free vs Plus vs Pro Plans](https://www.activepieces.com/blog/genspark-pricing-2026-free-vs-plus-vs-pro-plans) ## References - [Metronome](https://metronome.com/pricing-index/perplexityai)When a team treats the download as the purchase, it buys the wrong thing, or buys nothing and hits a usage wall in the middle of a task.