Social media scraping API

Strip the word "scraper" away and the job list is short. Almost every social data product needs the same handful of reads.

• —Profiles. Follower counts, bio, verification, and links, for a creator or a brand account.
• —Posts. Captions, media, timestamps, and engagement for a feed, a hashtag, or a single URL.
• —Comments. Comment threads and replies, with authors, for listening and moderation.
• —Search. Find accounts, posts, and hashtags by keyword before you read them.
• —Numbers you can trust. Likes, views, shares, and saves. And a clear signal when a number is hidden, not a zero.

1. 01Build a scraper. You write parsers against the private web and app interfaces of each platform. You buy proxies. You handle logins, blocks, and captchas. It works on day one, and then the platform ships a change. Every fix is your fix, and the engineer who wrote it is now on call for it.
2. 02Rent a scraper on a marketplace. You pick a listing on RapidAPI or a marketplace like Apify. It is fast to start and cheap for a prototype. In production you inherit the listing author's uptime, their update pace, and their billing tiers. Many listings share one backend, so several "alternatives" break together. We cover this on the RapidAPI vs direct providers page.
3. 03Call one API. You call one endpoint per read and get one response shape back, on every platform. Openhandle has 86 endpoints across Instagram, TikTok, and X today. Platform changes are absorbed on our side. You pay per answered request, after use, with a hard cap you set.

The failures are the same whether you built the scraper or rented it. They show up in the data, not in an error log.

• —Silent zeros. A metric the platform hid comes back as 0. Every average you compute from it is quietly wrong.
• —Empty 200s. A private or deleted account returns a success with blank fields. You cannot tell it apart from a real profile.
• —Expiring media links. Image and video URLs die within days. You must copy every file or lose it.
• —A new shape per platform. Instagram, TikTok, and X each come back different. Your code carries three parsers and three sets of edge cases.
• —Blocks at the worst time. Proxies get flagged during a campaign, a launch, or a payout run. That is when your users are watching.

• —Reads on request only. A customer asks for a profile or a post. We read it and answer. We do not crawl platforms on our own.
• —One schema. One schema across Instagram, TikTok, X, and Reddit. A profile is a profile and a post is a post, on every platform.
• —Null means hidden. If a platform hides a number, you get null. We never fill in a zero or a guess.
• —Media that stays up. Requested media is served from our CDN. Your link keeps working after the platform link expires.
• —Failed requests are free. Failed requests are free. You pay only when Openhandle returns an answer.
• —Pay afterwards. $0.0025 per request, down to $0.002 at volume. 30-day cache hits are free. Balance never expires. Unused balance is refundable.
• —Hard caps. Set a spend cap per key and per month. A cap is a stop, not an auto top-up.

Openhandle is read-only and public-only. We never like, follow, post, or message. We do not read private accounts, direct messages, or anything a logged-out visitor cannot see. Account owners can ask us to block their profile. The public data notice lists every limit.

Sign up, create a key, and make your first call. You get 100 free live requests and no card is needed. The quickstart has a working request in cURL, TypeScript, Go, and Python. Per-platform detail is on the Instagram, TikTok, and X pages, and every rate is on the pricing page.