Browse AI Task Limits and Credits Explained

In Browse AI, a task is one execution of a robot. A credit measures the work that execution uses, based on the data extracted and the target site’s cost classification. One run can therefore consume more than one credit. Recurring monitor checks and workflow-triggered executions count as tasks too.

To estimate usage, count how many pages your robots will check and how often, then account for extracted rows, text values, screenshots, and whether Browse AI classifies each site as standard or Premium. Published limits and plan details can change, so use your workspace’s Billing page for the plan and charges that apply to your account.

1. Tasks, robots, monitors, and credits

A robot is configured to perform a job; it is not itself a task quota. Each time the robot runs, Browse AI records an execution as a task. A monitor schedules repeated runs, and separate monitors can target different pages with the same robot. Browse AI’s API documentation gives daily monitoring of one page as approximately 30 tasks per month. Browse AI API documentation

Tasks and credits answer different questions:

Tasks: How many robot executions occur?
Credits: How much extraction or capture work do those executions consume?

Do not read a bulk-run capacity as a credit allowance. Browse AI documents bulk runs of up to 50,000 tasks at once. That describes how many tasks can be launched in a bulk operation, not how many credits your subscription includes.

2. How Browse AI calculates credits

Browse AI’s Help Center describes these rates, as of its March 13, 2026 article. They are vendor-published rates, not independent measurements. A task also has a minimum cost, so a small result can still consume the minimum.

Usage unit	Standard site	Premium site
Minimum per task	1 credit	2–10 credits
Rows extracted	10 rows per credit	10 rows cost 2–10 credits
Screenshots	1 screenshot per credit	1 screenshot costs 2–10 credits
Captured text values	10 values per credit	10 values cost 2–10 credits

Premium-site costs vary. Browse AI cites factors such as bot detection, CAPTCHA, complex or dynamic page structures, and access restrictions. Do not assume a particular site’s classification or an exact multiplier without checking the usage reported for your robot and account. Browse AI’s credit calculation guide

Examples of credit usage

50 rows from one standard-site list page: 50 × 0.1 credit per row = 5 credits.
A list plus product detail pages: extracting 100 product links costs 10 credits in the documented standard-site example. Visiting each of the 100 detail pages creates 100 separate tasks at the one-credit minimum, for 100 more credits. Total: 110 credits.
Monitoring 10 detail pages every three days: the documented standard-site example uses 10 credits per check, or about 100 credits per month.
Daily checks of one page: approximately 30 task executions per month; credit use depends on what each run captures and the site’s classification.

These are Browse AI’s examples, not guaranteed costs for every robot or website. A run that captures multiple data types or encounters a Premium site can use more than the minimum.

3. Estimate your monitoring allowance

List each monitored page. Count detail pages individually when the robot visits each one; a list page that reveals links does not make the detail-page visits part of the same task.
Calculate runs per month. Multiply pages by checks per page. For example, 10 pages checked every three days is roughly 100 page checks in a month.
Estimate data per run. Record expected rows, screenshots, and captured text values. Use the rates in the table as a starting point.
Include the task minimum and site range. Apply at least one credit per standard-site task, or the stated two-to-ten-credit minimum range for a Premium-site task.
Compare the estimate with actual usage. Run the robot, inspect usage reporting, and revise the estimate based on what it actually captures and how Browse AI classifies the site.
Leave room for extra runs. Manual reruns and workflow-triggered executions add tasks and can consume credits too.

A simple planning model is: monthly credits ≈ executions × estimated credits per execution. For monitoring, executions are approximately pages checked × checks per page per month. This is a planning estimate, not a promise of a fixed total: captured volume and site classification affect actual usage.

4. Published plan allowances and billing periods

Browse AI’s Help Center article dated March 13, 2026 describes these current plan configurations and says its plan explanation does not cover legacy plans. The listed choices are published configurations; they do not guarantee that every workspace can select every option.

Plan	Published credit configurations	Websites	Users
Free	50 monthly	2	3
Personal	2,000 monthly; 12,000 or 24,000 annually	5	3
Professional	5,000, 10,000, 20,000, or 30,000 monthly; 60,000, 120,000, 240,000, or 360,000 annually	10	10
Premium	600,000+ annually; custom terms	Custom	Custom

Monthly subscriptions have a monthly credit allowance. Browse AI says annual subscriptions provide the year’s credits upfront, without an additional monthly usage cap; the annual allocation is available during the subscription term. Unused credits do not roll over after the billing period ends, according to its credit guidance. Browse AI’s plan and pricing guide

Prices and selectable configurations can change. The public pricing page may show selected configurations and can differ from Help Center details, including add-on information. Check the pricing page and then the Billing view in your own workspace before purchasing. The Help Center describes added credits as one-time purchases for the plan cases it covers, with a 1,000-credit minimum, but published monthly add-on rates are not fully consistent across pages. Use the amount shown in Billing rather than assuming a universal top-up rate. Browse AI pricing

If your Billing page labels your subscription “Legacy,” the current-plan table may not describe its terms. Browse AI says legacy plans can be identified by that label in Billing. Review your account’s plan and charges there before estimating available credits. Browse AI plan guidance

5. Common estimation mistakes

Counting robots instead of executions: one robot can run repeatedly. Count each scheduled, manual, or triggered run as a task.
Assuming one task always costs one credit: one is the standard-site minimum. Data volume can raise usage, and Premium-site tasks have a higher minimum range.
Counting a list and all detail pages as one task: visiting individual detail pages creates separate tasks in Browse AI’s example.
Confusing bulk capacity with plan credits: the 50,000-task bulk-run capability does not grant 50,000 tasks’ worth of credits.
Assuming annual credits arrive monthly: Browse AI describes annual allocations as available upfront, with no additional monthly cap.
Assuming unused credits carry over: the credit guide says they do not roll over past the billing period.
Using a public add-on price as a guaranteed account price: published details differ in places. Check Billing for the charge applicable to your account.

6. Troubleshooting credit and task questions

Symptom	Likely cause	What to check or do
A run used more than one credit	Rows, screenshots, or captured text values added usage; the site may have a Premium classification.	Review the robot’s captured output and account usage. Compare it with the standard and Premium ranges, and avoid assuming a site class in advance.
Credits seem to disappear faster than task count suggests	Credits measure captured work and target-site cost, not just the number of runs.	Break usage down by robot and execution, then check rows, screenshots, text values, and site classification.
Monitoring uses more tasks than expected	Each page check is an execution, and multiple pages or more frequent schedules multiply runs.	Recalculate pages × checks per month, including separate detail-page visits and any manual or workflow-triggered runs.
The plan allowance does not match a public table	The workspace may have a legacy plan or an option not represented by the current summary.	Check the plan name, allowance, and charges in Billing. Treat the Help Center table as published configurations, not an account-specific quote.
An add-on price differs from an article or pricing page	Published add-on details are not fully consistent and can change.	Use the amount and minimum displayed in your workspace’s Billing page before purchasing.
Annual credits appear available all at once	Browse AI describes annual credits as front-loaded for the subscription term.	Plan usage across the full term; the Help Center says annual subscriptions do not have a separate monthly cap.

7. Performance, reliability, and cost planning

Task and credit limits are separate from how quickly a robot completes a run. The cited plan and credit materials do not provide a universal completion-time or reliability guarantee, so do not use task limits to infer throughput or a completion deadline.

For a reliable budget, start with a small representative set of pages, inspect actual extracted rows and values, and compare usage reporting with your estimate. Include retries or manual reruns in the budget if your workflow uses them. Site behavior can affect classification and captured output, so an estimate based on another site may not transfer.

Before choosing monthly versus annual billing, compare the allowance with your expected schedule and confirm the applicable account terms. Annual billing makes the full annual allocation available upfront according to Browse AI; it does not mean the credits roll over beyond the subscription period. Verify any add-on charge in Billing before relying on it as overflow capacity.

8. ScreenshotNeo as an alternative for screenshot capture

If your job is to capture rendered pages as images or PDFs rather than extract structured data with a robot, ScreenshotNeo is an alternative to try first. It is a website screenshot API and MCP server. It is not a Browse AI credit plan: it charges for clean screenshot captures under its own plans.

One GET request returns an image or PDF. For example, save a WebP screenshot with cURL:

curl -G "https://api.screenshotneo.com/v1/shot" -d access_key=YOUR_API_KEY --data-urlencode url=https://stripe.com -o shot.webp

See the ScreenshotNeo API documentation for request options, including output format and capture settings.

ScreenshotNeo removes known cookie and consent banners, newsletter popups, and chat widgets before capture; those steps can be turned off. Bot checks, CAPTCHAs, blank pages, timeouts, failed loads, and cache hits cost nothing, and the response identifies the page verdict and billing status in headers. AI agents can use its MCP server, which provides take_screenshot, get_page_info, and capture_pdf. The free plan includes 1,000 screenshots a month with no card; paid plans start at $5 for 3,000 screenshots. Every feature is on every plan.

9. FAQ

What does “task” mean in Browse AI?

One execution of a robot. A monitor’s scheduled check is an execution, as is a manual or workflow-triggered run.

Do credits roll over?

Browse AI’s credit guidance says unused credits do not roll over after the billing period ends.

Does the 50,000-task bulk run mean my plan includes 50,000 tasks?

No. It is the documented maximum tasks a bulk run can launch at once; plan credits are a separate allowance.

How many credits will my specific website use?

The public guidance gives standard and Premium ranges, but actual usage depends on the site’s classification and what the robot captures. Check usage for your own runs.

Where can I find my exact plan or add-on price?

Check Billing in your Browse AI workspace. This is especially important for legacy plans and add-on credit purchases.