The cheapest web scraping provider depends on your target pages, usable records and whether you can build your own extraction workflow. A low-priced API can be economical for simple HTML but much less generous once rendering, residential routing or protected domains consume extra credits. This guide compares ScrapingAnt, Apify, Novada, Scrape.do, ScraperAPI and Octoparse. See our broader best web scraping tools guide for more choices.
Prices are US dollars from public offers reviewed September 30, 2026. Credits, Actor usage, cloud processes and exported rows are different units. We did not run matched speed or success-rate tests. Ranking reflects pricing clarity and fit for the stated task.
Best flexible free platform
Apify
Hosted data workflows. Billing: resources and actor events.
Best multi-product starter pilot
Novada
Testing several scraping routes. Billing: successful results / product usage.
Current price snapshot
How to choose an affordable scraping provider
A low subscription helps only if the product fits the workflow.
Set a workload budget
Estimate URLs, retries, rendering, proxy mode and output size. Translate each provider’s units into a bill.
Separate access from extraction
A retrieval API may return HTML while parsing, deduplication and storage stay with you. Actors and visual tasks can bundle more steps.
Pilot difficult targets
Test authentication, localization, JavaScript and anti-bot handling on real targets. A general success claim cannot predict your yield.
Count engineering time
A visual plan can be cheaper for one-off work. An API may be cheaper at repeat scale when your team owns a pipeline.
For raw proxy traffic, compare the mobile proxy guide separately. Proxy bandwidth and successful API results are different billing units.
Quick comparison: six budget options
Entry price, free access and billing units side by side.
| Provider | Paid entry | Free option | Entry allowance | Billing unit | Best fit |
|---|---|---|---|---|---|
| ScrapingAnt | $19/mo | 10,000 credits monthly | 100,000 credits | API credits | Small developer projects |
| Apify | $19/mo + usage | $5 monthly platform usage | $19 prepaid usage | Resources and Actor events | Hosted data workflows |
| Novada | $20 starter | Confirm signup offer | Shared credit balance | Successful results / product usage | Testing several scraping routes |
| Scrape.do | $29/mo | 1,000 credits monthly | 250,000 Hobby credits | Successful-response credits | Teams needing several proxy modes |
| ScraperAPI | $49/mo | 5,000-credit trial; FAQ says 1,000 free | 100,000 Hobby credits | Core credits; Pipeline separate | Scheduled batches and structured endpoints |
| Octoparse | $83/mo monthly | 10 local tasks | 100 Standard tasks | Subscription and add-ons | Visual extraction |
Novada's starter is a purchase described in its pricing FAQ. Apify's $19 includes prepaid use but does not cap all possible spending at $19.
Pricing analysis and charts
Published prices and arithmetic examples, not equivalent workloads or measured performance.
Paid entry commitment
USD; Novada is a starter purchase.
What 100,000 credits could buy
ScrapingAnt Enthusiast, one mode at a time; theoretical requests.
Credit cost by request mode
ScrapingAnt's four alternative published modes, shown as relative segments.
The denominator is 161, the sum of the four example costs. This shows relative cost, not a workload mix. AI extraction can add more credits.
Compare cost per accepted record
Use the same permitted URLs, fields and quality checks for each pilot. Divide subscription + extra credits + proxy or Actor charges + engineering time by records passing those checks. For example, $19 provider spend and $31 staff time yielding 2,500 valid rows is $0.02 per accepted row. This is an illustration, not a provider benchmark.
Detailed provider reviews
What each service can replace, and what can change the bill.
ScrapingAnt
Best low-cost paid APIEnthusiast includes 100,000 credits, but request type determines how far they go. A simple datacenter request costs one credit, a rendered datacenter request ten, a simple residential request 25, and a rendered residential request 125. HTML, Markdown and AI extraction use the same pool; AI extraction adds a character-based charge. This is a good entry when your team can own scheduling, quality checks and storage.
Strengths
- Recurring free credits
- Managed browser and proxy routing
- HTML, Markdown and AI extraction
- Failed requests described as non-billable
Watch-outs
- Rendered residential requests use credits quickly
- No verified native dataset store or scheduler
- Concurrency marketing differs from error docs
Apify
Best flexible free platformStarter costs $19 monthly and includes $19 of prepaid platform usage, with additional use charged separately. Ready-made or custom Actors can cover extraction, scheduling and storage. Compute, residential proxies, transfer and storage affect the bill; some Store Actors also charge per event. Inspect the specific listing and cap a sample run before forecasting monthly cost.
Strengths
- $5 monthly free usage
- Schedules, datasets, API and webhooks
- Ready-made and custom Actors
- Complete hosted workflow option
Watch-outs
- Actor fees vary by listing
- Residential traffic is metered
- Monthly prepaid usage does not roll over
Novada
Best multi-product starter pilotNovada offers Web Unblocker for raw HTML, Scraper API for structured output and Browser API for interactive pages. Its scraping-pricing FAQ describes a $20 Starter Plan and one balance for these tools. Purchased credits normally expire after 30 days. The FAQ says there is no completely free trial while the homepage promotes one, so confirm the signup offer. Its advertised low per-1,000-result rate is not a fixed price for every product or target.
Strengths
- Shared balance across scraping tools
- Structured JSON or CSV for supported targets
- Documented success-based billing
- Several routes for different page types
Watch-outs
- No one public fixed request allowance to compare directly
- Purchased credits normally expire after 30 days
- Free-trial messaging varies by page
Scrape.do
Best broad API feature accessHobby lists 250,000 credits and ten concurrent requests. Standard datacenter retrieval costs one credit, rendered datacenter five, residential or mobile ten, and rendered residential or mobile 25. Some named domains cost more. Its async documentation gives different concurrency figures, so load-test sustained throughput before committing.
Strengths
- Core modes available on free tier
- Rendering and geographic routing
- Async jobs and webhooks
- Successful-response billing model
Watch-outs
- Special-domain multipliers
- Concurrency documentation conflicts
- Parsing and storage usually stay with you
ScraperAPI
Best crawler and pipeline suiteHobby costs $49 monthly for 100,000 credits and 20 concurrent threads. One credit buys a normal core API page; rendering, premium routing and named domains cost more. DataPipeline has its own credit table. The service also offers a Crawler and supported structured endpoints. Hobby and Startup geotargeting is limited to US/EU regions in the current plan table.
Strengths
- Crawler, DataPipeline and structured endpoints
- Rotating proxies and anti-bot handling
- URL cost estimator and spending cap
- Clear thread limits
Watch-outs
- Entry is $49, not the former $29
- Trial and free-plan wording differ
- Mode and Pipeline costs vary
Octoparse
Best no-code desktop builderThe free desktop plan has ten local tasks, up to 10,000 rows per export and 50,000 exported rows monthly. Standard costs $83 month to month, or an advertised $69 monthly equivalent with annual billing. Paid cloud execution, scheduling and rotation come with up to three concurrent cloud processes on the pricing page. Residential bandwidth and other add-ons can cost extra.
Strengths
- Point-and-click selection and pagination
- Useful local free plan
- Paid cloud scheduling
- Several export formats
Watch-outs
- Cloud needs a paid plan
- Higher entry commitment than APIs
- Add-on costs need review
Complete entry-tier pricing table
Public monthly terms unless noted; promotions, taxes and annual terms can differ.
| Provider | Free / trial | First paid option | Next public tier | Key condition |
|---|---|---|---|---|
| ScrapingAnt | 10K monthly credits | $19 / 100K credits | $49 / 500K credits | Mode and AI multipliers |
| Apify | $5 monthly usage | $19 Starter + PAYG | $199 Scale + PAYG | Resources and Actor events |
| Novada | Trial offer differs by page | $20 Starter (FAQ) | Plan-dependent | Shared credits; 30-day expiry |
| Scrape.do | 1K monthly credits | $29 / 250K credits | $99 / 1.25M credits | Mode and domain multipliers |
| ScraperAPI | 7-day 5K trial; FAQ says 1K free | $49 / 100K credits | $149 / 1M credits | Domain/mode; Pipeline separate |
| Octoparse | 10 local tasks | $83 Standard monthly | $299 Professional monthly | Cloud processes and add-ons |
Octoparse advertises $69 per month with annual billing for Standard; ScraperAPI advertises 10% off annually. Confirm live checkout terms.
What changes the real monthly cost?
A fair comparison starts with the same target set and an itemized bill.
1. Target difficulty and request mode
Static public pages may need only a basic datacenter request. Dynamic pages can require browser rendering; protected pages may need residential routing, retries or a specialized endpoint. On a 100,000-credit ScrapingAnt plan, switching every request from basic datacenter to rendered residential changes theoretical capacity from 100,000 to 800. Scrape.do and ScraperAPI use different multipliers, so estimate each independently.
2. Output and workflow charges
HTML access is only the first stage. Apify can charge for Actor events, compute, proxy traffic, storage and transfer. ScraperAPI DataPipeline has a separate credit schedule from its core API. Octoparse may add template, CAPTCHA or residential bandwidth charges. Novada's shared credits can move among products, but a Browser API job is not priced like a simple structured result.
3. Failed and unusable results
A request can be billed as successful yet return an empty field, duplicate record, stale listing or irrelevant page. Scrape.do's published success billing includes some HTTP statuses that are not useful data, and Novada notes an empty target page can still be billed if the request completed. Track the provider's billable event separately from your own accepted-record count.
4. Timing and unused credits
A monthly allowance may expire before an occasional project uses it. Apify's monthly prepaid usage does not roll over, ScraperAPI resets unused subscription credits, and Novada's purchased scraping credits typically expire after 30 days. For irregular research, a free local tool or carefully capped usage plan may cost less than a large nominal credit bundle.
A practical pilot worksheet
Record target URL, request mode, provider-reported credits, HTTP result, extracted fields, accepted record count, retries and time spent fixing selectors. Run a small sample of easy and hard targets. Project the monthly total using your observed mode mix, then add a buffer for target changes. The best value is the lowest dependable cost per accepted record for that specific task.
Features comparison matrix
Availability does not mean every feature is included at the entry price.
| Feature | ScrapingAnt | Apify | Novada | Scrape.do | ScraperAPI | Octoparse |
|---|---|---|---|---|---|---|
| Recurring free use | 10K credits | $5 usage | Offer varies | 1K credits | FAQ: 1K | Local plan |
| JavaScript pages | Managed browser | Browser Actors | Browser API | Rendering | Rendering | Desktop/cloud |
| Scheduled jobs | External tool | Yes | Confirm by product | Async jobs | DataPipeline | Paid cloud |
| Stored datasets | Bring your own | Yes | Product-specific | Bring your own | Pipeline output | Export |
| Structured extraction | AI endpoint | Actor-dependent | Scraper API | Ready APIs | Supported endpoints | Visual selectors |
Best use cases by provider
Match the billing unit and workflow to the job.
Small custom API integration
Start with ScrapingAnt’s free credits and measure your mode mix.
ScrapingAntRecurring data workflows
Use an Apify Actor with schedules and datasets; inspect its listing fees.
ApifyStructured data plus browser access
Pilot Novada’s shared balance across its three scraping tools.
NovadaSeveral proxy and rendering modes
Scrape.do offers several modes; check target-specific costs.
Scrape.doScheduled batches and structured endpoints
Evaluate ScraperAPI DataPipeline separately from the core API.
ScraperAPIFor a deeper API-versus-visual comparison, see ScraperAPI vs Octoparse.
Frequently asked questions
Questions that change a budget estimate.
How we compared these providers
We reviewed public pricing and product documentation on September 30, 2026. We compared entry payments, included units, multipliers, free access and limits. Charts use published prices or arithmetic from published credit costs. We did not purchase accounts, run matched jobs, verify IP inventories or measure success rates. Confirm checkout terms and test representative targets before buying.
For implementation context, the MDN Fetch API guide explains ordinary web requests and responses. Managed scraping adds provider-specific routing, browser execution or extraction.
Choose a pilot that matches your workflow
Use representative targets and a spending cap, then measure usable output before increasing volume.
Disclosure: FreePolicyTools may earn a commission from sponsored provider links. Rankings reflect the stated criteria; prices and offers may change.