Head to head · Extraction · October 2026 research run

Qovia vs Scrapling

Scrapling scores 68.5 (B) on agent readiness against Qovia's 48.1 (D), and leads in 4 of 7 scored categories. Qovia leads on schema & documentation, agent ergonomics and security & auth. Both do extraction.

Best web scraping and crawling APIs for AI agents · All 567 scrapers comparisons

Which one, for what

Qovia D

Best for Agents that need structured public data from Chinese platforms (Douyin, Xiaohongshu, Weibo, Bilibili, Kuaishou, Zhihu, WeChat official accounts, Taobao and JD reviews) or the Xingtu and Pugongying creator marketplaces, with TikTok, YouTube, Instagram, X and Reddit beside them, paid per successful call from a prepaid balance.

Ahead on

  • Schema & documentation, 86 against 81
  • Agent ergonomics, 77 against 71
  • Security & auth, 60 against 39

Also in its favour

  • A hosted endpoint, with nothing to install

Watch for

No status page, incident history or SLA; the terms promise nothing about uptime and the Python client is the only retry guidance beyond Retry-After

Scrapling B

Best for Agents that need to fetch and drive pages locally, including JavaScript-heavy or bot-protected ones, without a hosted scraping bill.

Ahead on

  • Reliability, 65 against 40
  • Payments & pricing, 90 against 70
  • Maintenance & community, 93 against 77
  • Transparency & trust, 52 against 30

Also in its favour

  • No key needed to call it
  • Runs on your own machine
  • Open source
  • No incidents deducted, where Qovia loses 15 points for them

Watch for

Version 0.5.1 with a Beta classifier. v0.4.15, v0.5 and v0.5.1 each carried breaking changes, the last in a patch release

Why the edge

Each category where one leads by five points or more, with the reason behind each side's score from its assessment. The numbers are in the table below; this is what they stand for.

Reliability Scrapling +25

Qovia 40

Hosted reading, for the REST API with the MCP server described. No public status page was found. https://qovia.ai/status/ answers 404, status.qovia.ai does not resolve and neither the docs nor the footer links one (0). No readable incident history, so 5 by the rubric (5). Limits are published with numbers on the quickstart's rate-limits section, 60 requests a minute per account before the first top-up and 600 after on a sliding 60-second window, with X-RateLimit-Limit and X-RateLimit-Remaining headers, a 64 KiB body cap, an empty-call throttle (50 calls in 10 minutes with over 80 per cent empty, not found or rejected) and a per-second cap per endpoint without a number (15). Retry-After is declared on too_many_auth_failures, rate_limited, too_many_unbilled, endpoint_busy and upstream_busy, the error table marks the retryable codes, the docs say a 503 endpoint_unavailable carries none and will not clear, and the Python client retries 429 and 503 honouring the header. Every operation is a read, including POST /v1/search, and a failed or timed-out call is refunded, so idempotency keys are not needed (15). No SLA. The terms summary says nothing about uptime and the pricing FAQ excludes outages, rate limits and platform changes from refunds; the 11 September 2026 archived home page advertised a 99.95 per cent enterprise SLA that the current pages have dropped (0). The endpoints are sold and labelled Live on the docs and coverage pages, but every one of the 143 carries x-qovia-status beta in the OpenAPI file, the API version is 0.1.0 and the MCP server reports 0.1.0 (5 of 10). 0+5+15+15+0+5 = 40.

Scrapling 65

Read on the local-software lines, since the MCP server runs on the user's machine over stdio or a self-run HTTP port. Installs from PyPI (scrapling[ai]) or Docker (pyd4vinci/scrapling, ghcr.io/d4vinci/scrapling), with Python 3.10 or later and 3.10 to 3.13 in the classifiers (20). tests.yml runs tox on Python 3.10 to 3.13 on macOS for pushes and pull requests to main and dev, and the repository holds about 1,390 test functions, 19 test files for the MCP tools among them. Run results on GitHub couldn't be read on 18 November, so passing on main is unchecked (15 of 25). Five issues were open on 18 November 2026 against 104,200 stars, the oldest from 24 October 2025, and two of them had a reply. The v0.5.1 notes credit fixes from outside contributors and close reported issues such as #540 (20 of 25). Breaking changes carry a warning and a migration guide in the changelog, but v0.5.1 broke remote browser connections in a patch release (10 of 15). Version 0.5.1 with a Development Status 4 Beta classifier (0). 20 + 15 + 20 + 10 + 0 = 65.

Transparency & trust Scrapling +22

Qovia 30

Closed service. The Terms of Service is one paragraph that names no legal entity, address or liability terms, says Delaware law governs and that the page is a summary with the signed agreement governing; the repository's client is MIT. Between the closed and unclear bands (7 of 30). The privacy policy is one paragraph. Account information is used to deliver and bill, personal information is not sold, requests and results are not resold, deletion requests go to the support mailbox, and the agreement governs where it differs. No retention periods, no subprocessors, no DPA, and no statement about the cache the prov.source flag implies; the SECURITY.md repeats the same points (8 of 30). A 410 endpoint_disabled code exists for retired endpoints and the changelog is dated, but there is no deprecation policy or notice period and nothing is marked deprecated (5 of 20). Stripe is named on the pricing page as the card and Alipay processor; no other provider, data location or transfer basis is stated (3 of 20). 7+8+5+3 = 23.

Scrapling 52

BSD-3-Clause in LICENSE and the PyPI classifiers (30). No telemetry or usage reporting found in the source. Page traffic goes to the target sites, with optional DNS-over-HTTPS through Cloudflare. There is no privacy statement for the package (15). No deprecation policy. Breaking changes are announced in the release notes with migration guides for v0.5 and the remote-session change (8). Nothing is collected that we could find, but there is no explicit statement on telemetry (12). 30 + 15 + 8 + 12 = 65.

Security & auth Qovia +21

Qovia 60

API keys with scopes chosen at creation (every endpoint, search, every raw endpoint or one of 24 platform groups), an optional expiry, revocation in the console and hashed storage; the docs say the key is read from the Authorization or x-api-key header only, never the URL or a cookie, and 30 failed authentications from one IP in a minute are throttled. No OAuth and no rotation feature beyond creating another key (24 of 30). The whole surface is read-only retrieval, scopes restrict a key to platforms, the two MCP catalogue tools need no key, and the docs and SECURITY.md tell users to give AI tools a separate scoped key (17 of 20). Every response is third-party platform text, and the only related statement is that error messages never contain the data source's own text; nothing on treating returned content as untrusted (3 of 15). request_id is the usage record ID on every answer, the console lists every charge in a usage log, X-Qovia-Charged and X-Qovia-Balance headers report each call, and GET /v1/account returns the key's scopes and balance at no charge; no exportable audit log or per-call endpoint was found (11 of 15). No security.txt on qovia.ai or api.qovia.ai (404), no bug bounty, SOC 2 or ISO 27001; the repository's SECURITY.md names a Telegram handle and a mailbox for reports covering both hosts (5 of 20). 24+17+3+11+5 = 60.

Scrapling 39

Read as local software with an optional HTTP transport. stdio needs no credential. Over HTTP the server refuses to start without a token or an explicit --no-auth, and accepts one shared bearer token with no scopes, rotated by restarting (15). Annotations let a client gate the destructive tools, and follow_redirects defaults to safe, which blocks redirects to private addresses. A direct request to localhost or a private address is not refused, and there is no read-only mode (10). Hidden elements, aria-hidden content, zero-width characters, comments and template tags are stripped by default before page text reaches the model, as documented in the MCP guide. The vendor's claim, untested by us (10). Browser sessions record network history the agent and operator can read back. No call log (4). No SECURITY.md, disclosure address or bug bounty in the repository, and GitHub's advisory page couldn't be read (0). 15 + 10 + 10 + 4 + 0 = 39.

Payments & pricing Scrapling +20

Qovia 70

No x402, MPP or L402 on Qovia's own endpoints; USDT on TRC-20 is a top-up method from a console session, not a machine payment protocol. The vendor's five Actors on the Apify Store under the profile qovia (Qovia Data, the same Telegram handle and mailbox as the site), such as the Xingtu Actor at $0.01 to $0.06 a section, can be paid through Apify, whose prepaid token is sold over x402, so partial credit for a third-party route on part of the catalogue (10 of 40). Per-call prices are published without a login, per endpoint in the OpenAPI file, llms.txt and the docs, search at $5 and raw endpoints from $0.5 to $19.5 per 1,000, with the tier rule and payment fees stated (20). 50 trial calls for a new account, counted only when a call returns data; the quickstart says to top up when they run out and the docs ask for no card at sign-up, though no account was created to confirm it (20). A person signs up in a browser and creates a key; the MCP catalogue tools work without a key but data calls do not, and the Apify route reaches four platforms without a Qovia account (5 of 20). 10+20+20+5 = 55. Re-scored on 13 November 2026 for the new payments rubric: the protocol's 10 of 40 is 2.5 of 10, and the other lines' 45 is 67.5 at 30 each, so 55 is 70.

Scrapling 90

Free, self-hosted software under BSD-3-Clause with nothing to buy, read with the self-hosted rule. No payment protocol (0), nothing to price (30), free (30), no signup or key (30). 0 + 30 + 30 + 30 = 90.

Maintenance & community Scrapling +16

Qovia 77

The newest changes are the registry entry ai.qovia/qovia 0.1.1 and the repository's tools snapshot and plugin manifests on 28 October 2026, after the Python client 0.1.1 on PyPI on 27 October; the repository was pushed on 1 November (30). Eight changelog entries on 21 and 22 October, two registry versions (22 and 28 October) and two PyPI releases (25 and 27 October) are more than three dated entries in 90 days, all within ten days (20). Closed service with a public changelog. Support is Telegram, WhatsApp, WeChat and one mailbox, with a one-business-day reply promised for email; GitHub issues are enabled on the repository with none opened, and no forum, Discord or community was found (7 of 15). ai.qovia/qovia is in the official MCP registry under a DNS-verified reverse-DNS namespace at 0.1.1, marked latest, and the Python client is an official package on PyPI (15). The client is standard-library Python with a test file, MIT, no dependencies and no CI workflow in the repository tree; the account's other nine repositories are forks of lists and directories (5 of 10). 30+20+7+15+5 = 77.

Scrapling 93

v0.5.1 tagged and released on 4 November 2026, 14 days before 18 November (30). v0.4.15 (23 August), v0.5 (17 October in the changelog, tagged 18 October) and v0.5.1 (4 November), three in the 90 days (20). The v0.5.1 notes credit about a dozen pull requests from outside contributors. Five issues and 18 pull requests were open on 18 November, and the issue template points to GitHub Discussions and a Discord (20 of 25). The official MCP registry lists io.github.D4Vinci/Scrapling as active, with versions from 0.4.1 (27 February 2026) to 0.5.1 (4 November 2026) (15). Tests and code-quality workflows, PyPI trusted publishing, and current dependencies such as Playwright 1.62. Run results unread (8 of 10). 30 + 20 + 20 + 15 + 8 = 93.

Agent ergonomics Qovia +6

Qovia 77

API reading. Responses come in fixed page sizes per endpoint, from about 5 comments on Xiaohongshu to 100 Fliggy reviews, with limit only on Bilibili danmaku and no field selection; the record shape is compact and shared across posts, notes and videos. The MCP server loads four tools by default, keeps the catalogue out of context until search_endpoints and get_endpoint are called, and ?tools= selects platforms or everything, which on the MCP reading would score 25 (18 of 25). Cursor pagination with has_more on most lists, search_id on Douyin and Zhihu search, page numbers on Weibo, Xiaohongshu and Bilibili, sort options on several, and the cross-platform search filters by platform, content type, sentiment and whole-day date windows up to 90 days (17 of 20). One error object with 20 codes, request_id on every answer, the retryable set marked, refunded codes separated from never-charged ones, and MCP tool errors carrying error.code, the HTTP status and retry_after_seconds (19 of 20). All calls are reads, failed and empty calls are refunded, Retry-After is declared and the Python client retries; no readOnlyHint or other annotation appears in the vendor's 28 October tools/list snapshot (14 of 20). One required parameter on 117 of 144 operations and defaults set on the server, but the only official SDK is the Python client; Node and curl are examples (9 of 15). 18+17+19+14+9 = 77.

Scrapling 71

17 tools, and v0.5 and v0.5.1 both shortened the descriptions. No toolsets or read-only subset (15). css_selector narrows fetch output, extraction can be Markdown, HTML, text or an ARIA snapshot with search, main_content_only is on by default, and network history is paged with a cursor and a limit of 1 to 200 (18). Failures from bad input, the browser and curl come back as tool errors with the message, which the v0.5 notes say were made clearer. No error codes an agent can branch on (12). Every tool carries annotations, read-only for fetches and lists and destructive for actions, JavaScript and state files. HTTP requests retry three times by default with a one-second wait (17). Only url is required for a fetch, with schema defaults. One language, Python, plus a CLI and Docker (9). 15 + 18 + 12 + 17 + 9 = 71.

Score by category

CategoryWeightQoviaScraplingEdge
Reliability16%4065Scrapling +25
Performance10%pendingpendingpendingnot scored in this run
Schema & documentation13%8681Qovia +5
Agent ergonomics13%7771Qovia +6
Security & auth14%6039Qovia +21
Payments & pricing10%7090Scrapling +20
Task success10%pendingpendingpendingnot scored in this run
Maintenance & community7%7793Scrapling +16
Transparency & trust7%3052Scrapling +22
Negative events≤15-150
TotalD 48.1/100B 68.5/100

Facts side by side

FactQoviaScrapling
KindHTTP APIMCP server
VendorQoviaKarim Shoair
Hosted endpointhttps://api.qovia.aino (local only)
TransportsHTTP, Streamable HTTPstdio, Streamable HTTP
AuthAPI keyNone
PricingPay per useFree
x402nono
LicenceProprietary hosted service under a one-paragraph Terms of Service summary that defers to a signed agreement. The qovia/qovia-mcp repository (setup guide, examples, Python client) and the qovia package on PyPI are MIT; the service's source is not publicBSD-3-Clause
Tools exposed14117
Read-only variant documentednono
llms.txtyesno
MCP registryai.qovia/qoviaio.github.D4Vinci/Scrapling
Last release2026-10-282026-11-04
Terms last updatednot read yetno document linked
Privacy policy last updatednot read yetno document linked
Popularity0 stars104k stars

How the scores have moved

Verdicts

Qovia

An OpenAPI 3.1 file with 144 operations, one envelope and 20 error codes, per-endpoint prices with failed and empty calls refunded, scoped keys and a four-tool MCP server that keeps the catalogue out of context. No status page, no legal entity, one-paragraph terms and privacy pages, and a product ten days old at the check.

Scrapling

Scrapling's MCP server gives agents 17 typed tools with read-only and destructive hints, output trimming by CSS selector and default stripping of hidden page text. It is pre-1.0 (0.5.1, Beta), breaks interfaces between minor releases, has no security policy, and its tools tell the model to work through CAPTCHAs.

Before you call either

Qovia

  1. Send Authorization: Bearer sk-qovia-... to https://api.qovia.ai and check meta.empty on every HTTP 200; data may be null or an empty structure and nothing is charged for it.
  2. Branch on error.code, never on message. Retry only rate_limited, too_many_unbilled, endpoint_busy, upstream_busy and too_many_auth_failures after Retry-After; endpoint_unavailable has none and will not clear.
  3. Stay under 60 requests a minute (600 after the first top-up) across every key on the account, and keep empty or not-found calls under 80 per cent of the last 10 minutes or expect 429 too_many_unbilled.
  4. Pass the previous page's cursor (and search_id on Douyin and Zhihu search) back unchanged and stop when has_more is false; each page is a separate charge, and unknown parameters answer 400.
  5. For MCP, connect to https://api.qovia.ai/mcp over Streamable HTTP; call search_endpoints then get_endpoint before call_endpoint, or add ?tools=douyin,kuaishou for typed direct tools. Catalogue tools need no key.

Scrapling

  1. Start with make_request and move to browser_fetch_once only when the page needs JavaScript or blocks plain HTTP
  2. Pass css_selector on fetch tools to cut output. Snapshots need one match or no selector
  3. Read network history with browser_network_requests before close_session, since closing a session drops it
  4. Check the target site's terms and robots.txt yourself. The MCP tools don't read robots.txt, only the spiders do
  5. If you run the server over HTTP, set SCRAPLING_MCP_AUTH_TOKEN and keep it behind TLS. The token travels in clear over plain HTTP

Questions

Which is better for AI agents, Qovia or Scrapling?

Scrapling scores 68.5 (B) on agent readiness against Qovia's 48.1 (D), and leads in 4 of 7 scored categories. Qovia leads on schema & documentation, agent ergonomics and security & auth.

Do Qovia and Scrapling need an API key?

Qovia needs an API key. Scrapling needs no key.

Can an agent call Qovia and Scrapling without installing anything?

Qovia has a hosted endpoint at https://api.qovia.ai. Scrapling runs on your own machine, with no hosted endpoint listed.

Are Qovia and Scrapling open source?

No open-source release is listed for Qovia. Scrapling is open source (BSD-3-Clause).

Other comparisons with Qovia or Scrapling

Machine-readable

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.