Perplexity Pro, Max, Enterprise, and API Pricing: Which Budget Covers What

The API changed significantly in September 2026: Perplexity ended support for Sonar Chat Completions on September 27, steering developers to its Agent API, which combines third-party models with search and other tools. This guide compares the app plans, explains how Computer credits work, sets out the current API meters, and shows when each route makes sense. Facts come from Perplexity's pricing pages, help center and API documentation, read on October 5, 2026. Perplexity Pro, How credits work, API pricing, Migrate from Sonar.

The App Plans

Pro includes extended Pro Search, access to advanced models from OpenAI, Anthropic, Google, xAI and others, image and video generation, a limited number of Create files and apps queries every 30 days, higher upload limits and up to 50 file uploads per project. Max adds the highest model access, extended Create files and apps, Brain (a research-preview memory system for Computer), Max Assistant in the Comet browser with the highest weekly limit on browser-agent queries, early access to new features and priority support. Annual Max billing is available only on the web app. Sign-up bonuses expire 30 days after they are granted and are one per account.

How Computer Credits Work

Credits pay only for Computer, Perplexity's agent that carries out multi-step work across the web, files and connectors. Ordinary Ask searches do not use credits, and neither does Comet's Assistant in Search mode; Work mode in Comet runs Computer and does. Perplexity prices credits at 100 per dollar and publishes typical ranges: light tasks such as summarizing an earnings call use 100 to 350 credits ($1 to $3.50), complex reports 350 to 950, heavy work such as a branded pitch deck 875 to 2,275, and large projects such as dashboards with data collection 2,400 to 9,800 ($24 to $98).

Max's 10,000 monthly credits therefore cover roughly 10 to 28 complex reports, or a handful of large projects. Monthly credits do not roll over. When they run out, active tasks pause and new Computer tasks are blocked until credits refresh or you buy more. Purchased credits are used after bonus and monthly credits and last until used, expiring only after a year without using Computer. Auto-refill is off by default and triggers at 500 credits on Pro and 2,500 on Max, and every Max account has a monthly credit spending cap, $200 by default and adjustable up to $5,000.

The API Platform Today

Perplexity's API is billed separately from every subscription, and Perplexity notes that Max's model access applies within the Perplexity web app, not to programmatic access. Its API now has several surfaces:

Perplexity's own examples help with budgeting. An Agent API request using an OpenAI model with 500 input and 200 output tokens and one web search costs about $0.0059, and a low-intensity research preset run with 2,000 input tokens, 1,000 output tokens, one search and one fetch about $0.007. The Search API charges per successful request, and up to five queries in one request count as a single billable unit.

What Happened to Sonar

Sonar Chat Completions support ended on September 27, 2026. Synchronous and streaming Sonar requests keep working because Perplexity is reformulating them as Agent API requests, rolling out gradually by model, but asynchronous Sonar requests are no longer supported and must move to the Agent API's background mode. Perplexity recommends migrating existing Sonar integrations to the Agent API and using it for all new projects. The Agent API still offers Perplexity's Sonar model alongside models from OpenAI, Anthropic, Google, xAI, Z.AI, Moonshot AI and NVIDIA, adds presets from fast to xhigh intensity, and follows the Open Responses standard.

If your budget was built on old Sonar per-request fees and token prices, rebuild it on the Agent API's model rates plus tool calls.

App or API: Which Route Fits

The decision is about where the work happens, not which plan is "better". If a person asks questions, reads sources and builds reports inside Perplexity, buy Pro or Max. If software needs search-grounded answers, ranked web results or an agent loop inside your own product, use the API. Pro versus Max is a separate question about how much Computer and research capacity one person needs.

Some rough comparisons make the boundary clearer. One analyst doing daily research fits Pro at $20 a month, with occasional Computer tasks paid from purchased credits. An analyst running several large Computer projects a week fits Max, because 10,000 monthly credits are worth $100 of tasks and the plan adds the highest model access. A product that answers 10,000 user questions a month with one web search each would cost roughly $59 on the Agent API at Perplexity's example rate, or $50 for 10,000 raw Search API requests if your own model writes the answer. A team of 20 researchers needing SSO and organizational controls fits Enterprise Pro at $40 per seat, or $800 a month.

Computer Credits in Dollars

Translating Perplexity's published ranges into dollars and into Max's monthly allowance shows how quickly credits go:

Perplexity stresses that these are examples, not guarantees, and that failed tasks caused by an error on its side have their credits restored. A Pro subscriber running five complex reports a month would spend roughly $17.50 to $47.50 in purchased credits on top of the $20 plan, still well below Max. A subscriber running two mega projects a week could need $200 to $780 of credits a month, where Max's included allowance and higher spending cap become worthwhile.

Enterprise Plans

Organizations have two seat-based options. Enterprise Pro costs $40 per seat a month or $400 a year, with a discounted $30 a month or $300 a year for eligible educational institutions and nonprofits, and includes 500 Computer credits per member each month. Enterprise Max includes 15,000 credits per member each month; confirm its seat price with Perplexity before budgeting. On enterprise plans, admins can set individual caps, so a member can be blocked when they reach their own allocation even if the organization's pool still shows credits, and members can request more through their usage settings. Enterprise seats do not include API usage, which is billed through the API Platform.

Pro or Max?

Choose Pro if you mainly use Ask and Pro Search, need advanced models and file uploads, and run Computer only occasionally. Its lack of monthly Computer credits keeps the price low, and you can buy credits when needed. Choose Max if Computer is part of your daily work, if you hit Create files and apps limits, or if you want Brain, Max Assistant in Comet and priority support. The jump from $20 to $200 is justified only if you would otherwise spend more than about $100 a month on Computer credits or need Max-only features. For more detail, see Perplexity Pro vs Max and is Perplexity Max worth it for research.

Controlling Spend on Both Routes

In the app, keep auto-refill off unless you rely on Computer, and set Max's spending cap to a level you are comfortable with. Use Ask rather than Computer for simple questions, since Ask does not use credits, and check the Usage dropdown in each thread to see what a task cost. On the API, choose Fast Search where latency matters and depth does not, use the cheapest model that answers well, limit tool calls per request, and monitor cost per request in the API portal, which reports total cost on completed responses.

Final Check Before Choosing

Decide first whether the work happens in the app or in your software. For the app, estimate how many Computer tasks you run each month and convert them to credits at 100 per dollar; if that value is under $100, Pro plus occasional purchased credits is usually cheaper than Max. For the API, list each request's model tokens and tool calls, multiply by monthly volume, and add a margin for retries. Do not count on a subscription to cover API usage, and if you still call Sonar Chat Completions, plan your migration to the Agent API now.