Web ScrapingUpdated August 2026

Maxun Review 2026: Pricing, Features, Pros & Cons

Item: Maxun
Rating: 4.4
Author: AISO Tools

Maxun turns a website into a structured API by recording your clicks, and then makes that extractor callable from a REST API, an SDK, a CLI or an MCP server. It is open source, self-hostable, and managed from $19 a month on credits. The interesting decision is not whether it works — it is whether you want to own the browser fleet.

Quick Verdict

4.4/5

Overall Rating

$0 or $19+

Self-host vs managed

Credits

Billed by work, not seats

Best for: teams building datasets or competitor monitoring, and agent developers who need web data callable over MCP. Not for: anyone who needs a predictable flat bill, or who wants a vendor to absorb the legal and anti-bot risk of scraping.

What Is Maxun?

Maxun is an open-source, no-code platform for turning any website into a structured API. It combines four operations that are usually spread across separate products: crawling an entire site and extracting page contents, scraping pages into LLM-ready Markdown or HTML, searching the web to find pages relevant to a query, and extracting structured data — either by recording your own actions or by describing in natural language what you want.

The recorder is what makes it usable by people who are never going to write a Playwright script. You browse the site the way you normally would, and Maxun converts those clicks into a production-ready extractor. Alongside it sits an autonomous agent path that navigates complex sites from a natural-language instruction. Reliability features are built into the platform rather than bolted on afterwards: proxy rotation and fingerprinting in a stealth mode, and scheduled monitors that run hourly or daily to track changes on a target automatically.

The architectural decision that distinguishes it is that every surface hits the same engine. A visual recorder, a REST API, Python and JavaScript SDKs, a CLI, webhooks, drop-in Skills for Claude Code, and a native MCP server so agents can call and orchestrate scrapers directly — none of these is a reduced version of the others. A library of prebuilt Auto Robots covers common targets such as GitHub trending repositories, IMDb by genre, Polymarket prediction markets, regional Google Trends, YouTube search results and Airbnb listings. The project reports over 100,000,000 rows extracted and more than 14,000 GitHub stars, and positions itself explicitly as a transparent alternative to black-box proprietary scrapers.

Maxun Pros & Cons

✓ Pros

•One extraction engine sits behind every surface — visual recorder, REST API, Python and JavaScript SDKs, CLI and MCP server — so what you build by clicking is the same thing you later call from code
•The native MCP server is the differentiator in this category: an agent can call and orchestrate a scraper directly, with no glue service in between, and Claude Code gets drop-in Skills as well
•It is genuinely open source and self-hostable, which caps your cost at infrastructure and removes the vendor as a single point of failure on a pipeline you depend on
•The recorder path means a non-engineer can produce a working extractor by browsing the site normally — the people who need the data are usually not the people who write Playwright
•Four jobs that are normally four products are in one place: full-site crawl, page-to-Markdown scrape, web search for relevant pages, and structured extraction by recording or by natural language
•Credit billing charges for work done rather than per seat, so a team of eight analysts on Growth costs the same as one analyst running the same volume
•Stealth mode with proxy rotation and fingerprinting is built in rather than sold as an add-on, which is where a lot of scraping budgets quietly go
•Scheduled hourly or daily monitors plus webhooks make change-tracking a first-class use case rather than something you cron yourself around a one-shot API
•Prebuilt Auto Robots for common targets — GitHub trending, IMDb, Polymarket, Google Trends, YouTube search, Airbnb — mean the first useful result arrives in minutes, not after a build

✗ Cons

•Credit accounting makes cost genuinely hard to predict: a target that changes its markup or adds a bot challenge can burn credits on retries, and the tier you sized last month may not hold this month
•Starter at $19/month gives 3,500 credits and a single user with only basic stealth, which is a trial-shaped plan more than a working one for anything commercially important
•The jump from Starter to Growth is $19 to $99 — a 5x price step — and the credit allocation is where the real difference sits, so there is no gentle upgrade path
•Stealth mode is a moving target. Anti-bot defences evolve continuously, and no vendor in this category can promise a target stays reachable, whatever the pricing page implies
•Self-hosting is free of licence cost but not of work: you are running a browser fleet, and browsers are the least well-behaved thing you can put in a container
•AI extraction from a natural-language instruction is the most impressive demo and the least deterministic feature — for a pipeline that must not silently drift, the recorder is the safer path
•Scraping carries legal and terms-of-service exposure that no tool removes; open source shifts the responsibility further onto you, not away from you
•Reported figures like 100,000,000+ rows and 14,000+ GitHub stars are the project's own, and stars are a popularity signal rather than a reliability one

Maxun Pricing 2026

Free

Self-hosted

•Full open-source platform
•No credit metering
•Your own infrastructure
•You own the browser fleet
•No usage ceiling but your hardware

Teams with someone who already runs containers

Starter

$19/mo

•3,500 credits
•1 user
•Basic stealth
•API access and webhooks
•Scheduled runs

One person validating a single data source

Growth

$99/mo

•100,000 credits
•5 users
•Advanced stealth
•API access and webhooks
•Scheduled runs

A working monitoring or dataset pipeline

Pro

$399/mo

•350,000 credits
•10 users
•Priority support
•Advanced stealth
•Full API, SDK and MCP access

Data teams running scraping as infrastructure

A custom-priced Premium tier adds an SLA and managed onboarding. Plan contents as published in August 2026 — verify before purchase.

What the Subscription Price Leaves Out

Cost line	Price	What to know
Free trial	7 days, no card	Long enough to answer the only question that matters: does the recorder survive your specific target for a week without hand-holding?
Premium tier	Custom	Adds an SLA and managed onboarding. The moment a scrape is load-bearing for revenue, the SLA is the line item you are actually buying
Self-host infrastructure	~$20–100/mo	Headless browsers are memory-hungry. The free route's real floor is a machine big enough to run several concurrently, plus proxies if the target pushes back
Proxies at scale	Your provider's rate	Stealth mode is included, but heavy volume against defended targets usually ends in a residential proxy bill that no scraping tool absorbs for you

The pattern worth internalising: in scraping, the subscription is rarely the expensive part. Proxies, retries against defended targets and the engineering time to repair broken extractors are where the real budget goes.

Who Should Actually Use Maxun

Use it if: you are building datasets, price monitoring or competitor tracking and need the extraction to be repeatable rather than one-off; you are building agents that need live web data and want a deterministic extractor behind an MCP call rather than an improvised fetch; the person who knows which data matters is not the person who writes scrapers; or you want the option to self-host so a vendor cannot end your pipeline with a pricing change.

Skip it if: you need a flat, forecastable monthly bill, because credit metering against changing targets will not give you one; you want a single well-supported API call and have no non-technical user in the loop, where an API-first competitor is a simpler fit; you have no capacity to fix an extractor when a site redesigns; or your target is a platform with a real API, in which case scraping it is the wrong answer regardless of how good the tool is.

Frequently Asked Questions

How much does Maxun cost in 2026?

Maxun is open source and self-hostable for free. The managed product starts with a 7-day free trial that does not require a credit card, then Starter at $19/month for 3,500 credits and one user with basic stealth, Growth at $99/month for 100,000 credits and five users with advanced stealth, and Pro at $399/month for 350,000 credits, ten users and priority support. A custom-priced Premium tier adds an SLA and managed onboarding. All paid plans include API access, webhooks and scheduled runs, so the tiers differ on volume, seats and stealth quality rather than on locking core capabilities behind a paywall.

What is a credit, and how do I estimate my bill?

Credits meter extraction work rather than seats, which is the fair way to bill this category and also the hardest to forecast. The honest approach is to refuse to estimate from the pricing page and instead run your real target through the trial for several days, then multiply. Two things distort the number: retries against defended sites, and targets whose markup changes and quietly breaks an extractor into a loop. Budget above your measured usage, and set up alerting on run failures rather than on credit balance — by the time the balance moves, the pipeline has usually already been wrong for a day.

Does the MCP server actually matter, or is it a checkbox?

It matters more than most MCP integrations, because web data is the thing agents most often need and least reliably get. A native MCP server means an agent can call a scraper you already built and validated, with the extraction logic pinned, rather than improvising a fetch-and-parse each time and hallucinating structure. That distinction — deterministic extractor called by a non-deterministic agent — is the useful architecture. Drop-in Skills for Claude Code extend the same idea. If you are building agents that depend on live web data, this is a genuine reason to prefer Maxun over an API-only scraper.

Should I self-host or pay for the managed plans?

Self-host if scraping is a durable capability for you and someone owns the infrastructure — the licence cost is zero and the ceiling is your hardware. Pay if the data matters more than the pipeline does. Running a browser fleet is genuinely operational work: memory pressure, zombie processes, proxy rotation and version drift are all yours, and they surface at inconvenient times. The rule of thumb that holds up: if nobody on the team would notice a silently failing scraper within a day, you want the managed plan, because the failure mode of self-hosting is not downtime, it is stale data that still looks fine.

How does Maxun compare to Browse AI and Firecrawl?

Browse AI occupies the same no-code recorder territory and is the closer product comparison; Firecrawl is the API-first crawl-and-Markdown option that developers reach for when there is no non-technical user in the loop. Maxun's argument against both is surface coverage: it does the recorder, the crawl-to-Markdown, the search and the AI extraction from one engine, and exposes all of it over MCP. The argument against Maxun is maturity of the managed service — open-source-first projects usually have more capability and less polish than a commercial competitor at the same price. Trial the same target on two of them before committing; the comparison that decides it is always your specific site, not the feature grid.

Which plan should most teams start on?

Start on the 7-day trial or a local self-hosted instance, and point it at the single data source you actually need — not at a demo target. If the recorder holds for a week without intervention, the tool works for you. From there, Starter at $19 is only right for one person validating one source; most teams that get value land on Growth at $99, because 3,500 credits disappear quickly once monitors run daily. Do not buy Pro on projected volume. Buy it when Growth's credits run out, which is a much more reliable signal than any estimate you can make in advance.