# RapidAPI Alternatives: 4 Profiles, One Author Schema (https://www.socialcrawl.dev/blog/rapidapi-alternatives)
> RapidAPI alternatives for Instagram, TikTok, and YouTube: one x-api-key. Live @mrbeast profiles, computed engagement rates, 17-source search for 33 credits.
Replacing RapidAPI's Instagram, TikTok, and YouTube scrapers is not another Hub listing. The RapidAPI alternatives that actually parse social data share one `x-api-key` and one Author / Post shape.
Send `x-api-key: $SOCIALCRAWL_API_KEY` at `https://www.socialcrawl.dev`. Profile and post calls return an `author` or `post` object plus `computed.engagement_rate`. The JSON in this post is trimmed from live 200s, not docs fiction.
An API *client* (Insomnia, Hoppscotch, Bruno, Apidog) sends HTTP requests. An API *marketplace* ([RapidAPI Hub](https://docs.rapidapi.com/docs/what-is-rapidapi)) is a directory of independently published APIs. A *unified social data API* returns one envelope and one object model. RapidAPI Hub is the middle bucket. This post is the third.
Eleven live GETs on 2026-09-02, 23:48–23:49 UTC, against public `@mrbeast` / `MrBeast`. Billed: 33 credits. Catalog on that run: 61 platforms, 523 endpoints, 29 social, billed 0.
## Why do developers look for RapidAPI alternatives?
Developers look for RapidAPI alternatives when they need Instagram, TikTok, and YouTube in *one parser*, not one billing dashboard. RapidAPI is a marketplace. One Rapid key authenticates to the Rapid Runtime; each listing still has its own `X-RapidAPI-Host`, plan, quota object, and JSON shape ([RapidAPI consumer guide](https://docs.rapidapi.com/docs/consumer-quick-start-guide), [keys docs](https://docs.rapidapi.com/docs/keys-and-key-rotation)). Social scrapers on the Hub are independent listings, not one Instagram product ([Hub search](https://rapidapi.com/search?term=instagram&sortBy=ByRelevance)).
Nokia acquired Rapid on 13 November 2024 ([Nokia press release](https://www.nokia.com/about-us/news/releases/2024/11/13/nokia-acquires-rapid-technology-and-rd-unit-to-strengthen-development-of-network-api-solutions-and-ecosystem/)). The Hub still markets “five million developers” connecting to “thousands of APIs” ([RapidAPI docs](https://docs.rapidapi.com/docs/what-is-rapidapi)). Nokia disclosed “thousands of active developers” and, in the TechCrunch write-up, “hundreds of APIs”; Rapid had been valued at $1B in 2022 with “as many as 4 million users” and “some 40,000 APIs” ([TechCrunch](https://techcrunch.com/2024/11/13/nokia-acquires-rapid-the-api-company-once-valued-at-1b/)). The two censuses disagree; quote both. A listing is not a production guarantee ([API7.ai](https://api7.ai/learning-center/api-101/rapidapi-discovering-and-connecting-to-apis)).
As of 2026-09-03, GitHub ranked #1 for this query and listed Insomnia and Hoppscotch as RapidAPI alternatives ([Awesome RapidAPI Alternatives](https://github.com/hervekom37/Awesome-RapidAPI-Alternatives)). That is like calling a REST client a payment processor. If you were comparing API *clients*, you already left the Hub. If you were comparing social data APIs, those READMEs never printed a response body. [Writing your own scraper](/blog/web-scraping) is the other fork — a different job from swapping Hub listings.
You start from one free catalog, not a search box of independently published scrapers.
```bash
curl -s -H "x-api-key: $SOCIALCRAWL_API_KEY" \
"https://www.socialcrawl.dev/v1/utility/endpoints"
```
`credits_used: 0`. `request_id: req-1FRjr5x1Q0irdMqp`. 136 ms. As of this 2026-09-02 run:
| Field | Live value |
| --- | --- |
| `data.stats.platforms` | 61 |
| `data.stats.endpoints` / `data.total` | 523 |
| `data.stats.social_platforms` | 29 |
| Methods | GET 506, POST 13, DELETE 3, PATCH 1 |
| Credit histogram (single-integer cost) | 1 cr × 260, 5 cr × 184, 10 cr × 24, 20 cr × 11, plus smaller buckets; 18 metered-range |
Those 61 / 523 / 29 social numbers will drift. Quote the catalog you just called, not a remembered docs figure. That catalog is the denominator for every RapidAPI alternatives comparison below.
What the same key returned:
| Capability | What came back | Credits |
| --- | --- | --- |
| One free catalog | 61 platforms, 523 endpoints, 29 social | 0 |
| One key, four platforms | `@mrbeast` Author objects on YouTube, X, TikTok, Instagram | 4 × 1 |
| Canonical Post + computed rate | TikTok 0.141637, YouTube 0.022298, Instagram photo `null` | 3 × 1 |
| Prism creator-card | `combined_followers` 778,847,326, `platforms_found` 4 | 5 |
| Prism lookup | YouTube watch URL → `/v1/youtube/video`, billed 1 | 1 |
| Universal search | 17 sources, 15 succeeded, 143 raw → 40 ranked | 20 |
### The 11 calls behind this post (33 credits, 2026-09-02 UTC)
Base `https://www.socialcrawl.dev`. Header `x-api-key`. Handle `mrbeast` / `MrBeast` because it exists on all four networks. Order: catalog → four profiles → three post lists → creator-card → URL lookup → `/v1/search/everywhere`.
Post-level `computed.engagement_rate` is `(likes + comments + shares) / views`, clamped to `[0, 1]`, and `null` when the divisor or numerator is missing. Shares that arrive as `null` do not enter the numerator as zero. Author-level rate is `likes_count / followers`. That is the wrong tool on mega-creators; TikTok returned `null` plus a warning rather than a 10.39 “engagement rate.” Use the post-level number. The same formula, measured across 1,269 posts, is in the [2026 engagement-rate benchmarks](/blog/social-media-engagement-rate-benchmarks-2026).
## How do you replace RapidAPI Instagram and TikTok scrapers with one schema?
Same header, four 1-credit profile GETs, handle `mrbeast` / `MrBeast`. On RapidAPI this is four listings, four hosts, four field names (`edge_followed_by.count` vs `followerCount` vs `statistics.subscriberCount` vs `public_metrics.followers_count`). Here the parser is `data.author.followers` four times. That is the scrape instagram api / tiktok scraper api job with one extractor.
```bash
curl -s -H "x-api-key: $SOCIALCRAWL_API_KEY" \
"https://www.socialcrawl.dev/v1/instagram/profile?handle=mrbeast"
```
Swap the path and the handle casing for the other three: `/v1/tiktok/profile?handle=mrbeast`, `/v1/youtube/channel?handle=MrBeast`, `/v1/twitter/profile?handle=MrBeast`.
| Platform | Endpoint | `author.id` | username | followers | posts_count | verified | `computed.engagement_rate` | latency | cr |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| YouTube | `/v1/youtube/channel` | `UCX6OQ3DkcsbYNE6H8uQQuVA` | `@mrbeast` | 515,000,000* | 999 | `null` | `null` | 312 ms | 1 |
| X | `/v1/twitter/profile` | `2455740283` | `MrBeast` | 34,974,299 | 7,997 | `true` | `null` | 1,337 ms | 1 |
| TikTok | `/v1/tiktok/profile` | `6614519312189947909` | `mrbeast` | 139,719,263 | 469 | `true` | `null`† | 1,494 ms | 1 |
| Instagram | `/v1/instagram/profile` | `2278169415` | `mrbeast` | 89,153,744 | 506 | `true` | `null` | 3,851 ms | 1 |
\* YouTube `author.ext.followers_approximate = true` — the platform's rounded subscriber figure, not an exact integer. † TikTok `_warnings`: author ratio exceeded 1.0 (raw 10.386896); lifetime likes 1,451,249,438. YouTube `verified` is `null`, not `false`. Do not write “unverified.”
Shared fields on every row: `id`, `username`, `display_name`, `bio`, `verified`, `followers`, `following`, `posts_count`, `likes_count`, `url`. Missing values are `null`, not guessed (YouTube `verified` / `following`, Instagram and X `likes_count`, TikTok `url`). Instagram's profile call was the slow leg this run, 3,851 ms. That is wall time, not an SLO.
Trimmed `data.author` from the TikTok call (`request_id: req-XJlNdak7HfAR9zwc`):
```json
{
"success": true,
"platform": "tiktok",
"endpoint": "/v1/tiktok/profile",
"data": {
"author": {
"id": "6614519312189947909",
"username": "mrbeast",
"display_name": "MrBeast",
"verified": true,
"followers": 139719263,
"posts_count": 469,
"likes_count": 1451249438,
"url": null
},
"computed": {
"engagement_rate": null
},
"_warnings": [
"computed.engagement_rate: author ratio exceeded 1.0 (raw: 10.386896); returned null. A lifetime likes/followers ratio is not a real engagement rate"
]
},
"credits_used": 1,
"request_id": "req-XJlNdak7HfAR9zwc"
}
```
The SocialCrawl API lists posts on the same `post.engagement.*` block. Three 1-credit lists, `data.dropped = 0` on all three.
```bash
curl -s -H "x-api-key: $SOCIALCRAWL_API_KEY" \
"https://www.socialcrawl.dev/v1/tiktok/profile/videos?handle=mrbeast"
```
TikTok `/v1/tiktok/profile/videos?handle=mrbeast` — 10 items, `has_more: true`. Post `7681056640212258078` (“Replying to @Joel_02 subscribe for a cookie”): views 1,517,508 · likes 196,159 · comments 12,074 · shares 6,703 · saves 11,286 · **`computed.engagement_rate` = 0.141637**. Arithmetic: `(196159 + 12074 + 6703) / 1517508 = 0.141637`. Next row on the same page: 0.1629.
YouTube `/v1/youtube/channel/videos?handle=MrBeast&includeExtras=true` — 30 items. Video `Qtl8lJwbd4g` (“Escape 100 Cops, Win $500,000”): views 88,570,486 · likes 1,889,935 · comments 85,000 · shares `null` · **0.022298**. Shares is `null` and does **not** enter the numerator as zero: `(1889935 + 85000) / 88570486 = 0.022298`.
Instagram `/v1/instagram/profile/posts?handle=mrbeast` — 12 items. Pinned photo `3946249480127175223` (`/p/DbD5P56DNo3/`): views `null` · likes 9,323,186 · comments 142,523 · rate **`null`**. The first two items were pinned grid posts, not Reels. Instagram does not expose a public view count on photo/carousel grid posts, so the API leaves `engagement.views` null and refuses to invent a rate. The null *is* the product.
Trimmed post and computed JSON from that TikTok video:
```json
{
"post": {
"id": "7681056640212258078",
"url": "https://www.tiktok.com/@mrbeast/video/7681056640212258078",
"content": {
"text": "Replying to @Joel_02 subscribe for a cookie 🍪"
},
"engagement": {
"views": 1517508,
"likes": 196159,
"comments": 12074,
"shares": 6703,
"saves": 11286
}
},
"computed": {
"engagement_rate": 0.141637
}
}
```
`computed.engagement_rate` is free. A RapidAPI Instagram scraper, a tiktok scraper api, and a YouTube Data API wrapper do not share `post.engagement.likes`. One client writes one extractor. If the job is comments rather than profiles, [page YouTube comments without a Hub listing](/blog/youtube-comment-scraper) is the sibling endpoint.
## What is a unified social media API, and why isn’t RapidAPI one?
A unified social media API returns the same object model (`dev.socialcrawl.post.v1` / `author.v1`) and the same envelope regardless of platform. RapidAPI’s “single API key” authenticates to N hosts; each listing keeps its payload ([RapidAPI docs](https://docs.rapidapi.com/docs/what-is-rapidapi)). Rapid unifies auth and metering, not data ([monetize tab](https://docs.rapidapi.com/docs/hub-listing-monetize-tab)). That is the split versus the Hub *and* versus pages that claim a `title/url/author/snippet` object without printing a body.
Every 200 in this post shares the envelope: `success`, `platform`, `endpoint`, `data`, `credits_used`, `request_id`, `cached`. One shape. A [monitoring pipeline](/blog/social-media-monitoring-api) can consume those canonical objects without a per-hub mapper.
Prism composites are the one-handle / one-URL proof. RapidAPI has no “look this handle up everywhere” product.
```bash
curl -s -H "x-api-key: $SOCIALCRAWL_API_KEY" \
"https://www.socialcrawl.dev/v1/prism/creator-card?handle=mrbeast"
```
5 credits flat, 4,697 ms, `request_id: req-q7T9StuQvRq0Rg84`. All four legs HTTP 200. TikTok 139,719,286 · Instagram 89,153,740 · YouTube 515,000,000* · X 34,974,300. `totals.combined_followers` = **778,847,326**. Check: `139719286 + 89153740 + 515000000 + 34974300 = 778847326`. `platforms_found` = 4. `any_verified` = true.
Billed 5 vs four standalone 1-credit calls. The extra credit is the composite’s own charge; the four legs internally report 1 cr each. `combined_followers` is a sum of four platform audiences, not unique people — YouTube is still rounded. Counts drifted a few units vs the standalone table over ~15 seconds (TikTok 139,719,263 → 139,719,286; X 34,974,299 → 34,974,300; Instagram 89,153,744 → 89,153,740). Quote composite totals for the one-call path.
Trimmed `totals` plus the slow leg (Instagram, 4,577 ms). Avatars dropped.
```json
{
"totals": {
"platforms_found": 4,
"combined_followers": 778847326,
"any_verified": true
},
"legs": [
{
"endpoint": "/v1/instagram/profile",
"status": 200,
"credits_used": 1,
"latency_ms": 4577,
"error": null
}
],
"credits_used": 5,
"request_id": "req-q7T9StuQvRq0Rg84"
}
```
Paste a URL when you do not want to pick a hub:
```bash
curl -s -H "x-api-key: $SOCIALCRAWL_API_KEY" \
"https://www.socialcrawl.dev/v1/prism/lookup?url=https://www.youtube.com/watch?v=Qtl8lJwbd4g"
```
1 credit (the resolved endpoint’s price; lookup adds 0). 482 ms. `request_id: req-M1gVaUv8DOWKbO0B`. `data.resolved.platform` = youtube · `endpoint` = `/v1/youtube/video` · `archetype` = Post. Same video as the list item above. Comments **85,171** (the list had 85,000 — different endpoint, ~15 seconds later). Rate **0.0223**: `(1889935 + 85171) / 88570486 = 0.0223`. RapidAPI makes you identify the platform, find the hub, and learn that hub’s video-id param. Lookup sniffs the URL. The join key for the rest of the unified schema is `url`.
4,697 ms for the card is this run’s wall time, not a product promise.
## What does `/v1/search/everywhere` return that RapidAPI Hub cannot?
RapidAPI cannot fan 17 social and research sources out of one listing. Instead of RapidAPI, one query:
```bash
curl -s -H "x-api-key: $SOCIALCRAWL_API_KEY" \
"https://www.socialcrawl.dev/v1/search/everywhere?query=instagram%20scraper%20api&lookback_days=60"
```
**20 credits**, 31,034 ms, `request_id: req-bFsi0kkLEz7SIgHN`, `cached: false`. The query string is `instagram scraper api` — the scrape instagram api job, searched rather than subscribed.
| Field | Live value |
| --- | --- |
| Sources called | **17** (polymarket, hackernews, github, linkedin, threads, tavily, youtube-hashtag, youtube, tiktok, instagram-hashtag, tiktok-hashtag, pinterest, perplexity, rumble, reddit, twitter-ai-search, instagram) |
| Succeeded / failed | **15** / **2** |
| Failures | `twitter-ai-search` “per-source timeout after 11000ms”; `instagram` “per-source timeout after 12000ms” |
| Coverage | **0.8824** (15/17) |
| Plan | intent `product` / freshness `balanced_recent` / cluster `workflow` · 6 subqueries |
| Raw → fused | **143** → **40** · 40 semantic clusters |
Do not round that to 17/17. Two sources timed out; the body says so in `sources_failed`. Threads, youtube-hashtag, and instagram-hashtag succeeded with **0 items**. Empty is not failed.
Raw items by source: github 30, tiktok 29, pinterest 18, youtube 16, hackernews 14, linkedin 10, tavily 8, reddit 7, perplexity 7, rumble 2, polymarket 1, tiktok-hashtag 1; threads / youtube-hashtag / instagram-hashtag = 0.
Top 5 fused results (verbatim titles, URLs, scores):
| # | Source | Title | `final_score` / `rerank_score` |
| --- | --- | --- | --- |
| 1 | hackernews | Instagram Scraper API (`https://instagramscraperapi.com`) | 62.54 / 95 |
| 2 | reddit | Complete Instagram Scraper Suite on Apify | 62.25 / 80 |
| 3 | tavily | Bright Data Instagram Scraper API docs | 57.14 / 88 |
| 4 | reddit | Instagram API Scraper — Pay per event \| 17K users | 56.04 / 86 |
| 5 | tavily | Instagram Scraper (`https://rapidapi.com/junioroangel/api/instagram-scraper`) | 54.19 / 84 |
Rank 5 is a live RapidAPI Hub URL, returned by this call, not planted. The marketplace this post is about still surfaces for “instagram scraper api.” The call that found it has no RapidAPI equivalent. It is a fused search hit, not an endorsement.
Reddit item #2 carried a real comment from `alOOshXL` (`truncated: false`): *“As someone who work in influencer agency and use social media scrappers daily Your price is very high especially for profile scrappers.”* Voice of customer, not a statistic.
Trimmed envelope: coverage, the two timeouts, and the rank-5 hit.
```json
{
"coverage": 0.8823529411764706,
"sources_failed": {
"twitter-ai-search": "per-source timeout after 11000ms",
"instagram": "per-source timeout after 12000ms"
},
"item": {
"source": "tavily",
"title": "Instagram Scraper",
"url": "https://rapidapi.com/junioroangel/api/instagram-scraper",
"final_score": 54.19228866240961,
"rerank_score": 84
},
"credits_used": 20,
"request_id": "req-bFsi0kkLEz7SIgHN"
}
```
Replacing a RapidAPI Instagram scraper is the small story. One ranked list across GitHub + HN + TikTok + YouTube + Reddit + Pinterest + LinkedIn + Perplexity is the large one. 31 seconds is this run, not an SLO.
## Is Apify a RapidAPI alternative, or a different product?
Three buckets. No ranking. The RapidAPI alternatives SERP treats Apify as a peer marketplace. It is a different product.
**RapidAPI Hub** is a marketplace of independently published APIs. One key, N hosts, N plans, N JSON shapes. Social scrapers are listings, not a product ([what RapidAPI is](https://docs.rapidapi.com/docs/what-is-rapidapi), [monetize tab](https://docs.rapidapi.com/docs/hub-listing-monetize-tab)).
**Apify Store** is a marketplace of Actors. Each Actor has its own input JSON and charge model ([run Actors](https://docs.apify.com/get-started/run-actors)). Instagram alone is four first-party Actors: `apify/instagram-scraper` from $1.50/1,000 results, post from $1.00/1,000, reel from $1.00/1,000, profile from $1.60/1,000; plus `clockworks/tiktok-scraper` from $1.70/1,000 and `streamers/youtube-scraper` from $2.40/1,000 videos (paid-plan floors as of 2026-09-03; free-plan rates are higher) ([Instagram scraper](https://apify.com/apify/instagram-scraper), [post](https://apify.com/apify/instagram-post-scraper), [reel](https://apify.com/apify/instagram-reel-scraper), [profile](https://apify.com/apify/instagram-profile-scraper), [TikTok](https://apify.com/clockworks/tiktok-scraper), [YouTube](https://apify.com/streamers/youtube-scraper)). Rental Actors stop taking new listings after 1 April 2026 and remaining rentals migrate to pay-per-usage on 1 October 2026 ([Store docs](https://docs.apify.com/actors/running/actors-in-store)). Apify is honest that its Store is scraper-focused. The apify vs rapidapi question is marketplace-of-Actors vs marketplace-of-APIs. Neither is a canonical Author / Post schema.
**A social data API** with one envelope and credits per call is this post’s subject. Bright Data is proxy/dataset. ScrapingBee is HTML fetch. If the job is [building a dashboard on one API instead of another SaaS seat](/blog/hootsuite-alternatives), that is a different SERP.
### What did these 11 calls cost?
| Call | `credits_used` |
| --- | --- |
| `/v1/utility/endpoints` | 0 |
| 4 × profile | 4 × 1 |
| 3 × post list | 3 × 1 |
| `prism/creator-card` | 5 |
| `prism/lookup` | 1 |
| `/v1/search/everywhere` | 20 |
| **Total** | **33** |
The same job on RapidAPI is four Hub subscriptions (Instagram + TikTok + YouTube + X scrapers) plus per-listing overage plus `X-RapidAPI-Host` per call. On Apify it is four Actors with four input schemas. Credit packs and the 100-credit signup grant live on [/pricing](/pricing). This post does not mint a fake Hub invoice.
YouTube Data API v3 is a different stack: GCP project, OAuth, 10,000 units/day ([Google](https://developers.google.com/youtube/v3/getting-started)). Marketplace scrapers sell themselves as the no-quota alternative. This API is neither Google’s official quota nor a Hub listing.
## How do you try the same 11 calls with 100 welcome credits?
Sign up, take the **100 welcome credits**, no card. That grant covers this 33-credit harvest three times over. One API key. Header `x-api-key`. [Quickstart](/docs/quickstart).
The first call is the 0-credit catalog from above, so you see 61 / 523 / 29 social before spending. Then one 1-credit profile from the Author table. After that: the [Explorer](/explorer) (see your data before writing a single line), [platform docs](/platforms), and [/pricing](/pricing).
Agents can hit the same surface over MCP. APIFreaks names MCP and never ships an install. Two real ways, both live: stdio with `npx socialcrawl-mcp` for Claude Desktop, Cursor, or VS Code; hosted Streamable HTTP at `https://mcp.socialcrawl.dev/mcp` for Claude Code or a cloud agent. Primer: [what an MCP server actually is](/blog/what-is-an-mcp-server). Worked tool calls: [real MCP tool calls against the same API](/blog/mcp-server-examples). We did not run MCP in this harvest, so there is no fake tool-call JSON here.
Captured envelopes always include `success`, `request_id`, and `credits_used`. 429s are retryable; see [rate limits](/docs/rate-limits). This post does not invent a 429 body. Anyone with a key can rerun the same 11 GETs. Counts will move (they are live). Envelope, Author / Post field names, and credit prices will not.
If you came here shopping RapidAPI alternatives, start with the catalog curl. It costs nothing.
## Frequently asked questions
### What is RapidAPI, and why do developers look for alternatives?
RapidAPI is a marketplace, not a unified schema. One key authenticates to the Rapid Runtime; each listing still has its own host, plan, quota object, and JSON shape ([docs](https://docs.rapidapi.com/docs/what-is-rapidapi)). Social scrapers are independent Hub listings. Developers look for alternatives to RapidAPI when Instagram + TikTok + YouTube need one parser.
### Is RapidAPI owned by Nokia?
Yes. Nokia acquired Rapid on 13 November 2024 ([Nokia press release](https://www.nokia.com/about-us/news/releases/2024/11/13/nokia-acquires-rapid-technology-and-rd-unit-to-strengthen-development-of-network-api-solutions-and-ecosystem/)). Terms were undisclosed. RapidAPI docs still say “five million developers” / “thousands of APIs”; Nokia disclosed “thousands of active developers,” and TechCrunch reported Rapid’s earlier “40,000 APIs” against Nokia’s “hundreds of APIs” ([TechCrunch](https://techcrunch.com/2024/11/13/nokia-acquires-rapid-the-api-company-once-valued-at-1b/)). The Hub is not shutting down. It is a marketplace under a new owner.
### How much does RapidAPI cost?
Per listing, not per platform. Providers mint BASIC / PRO / ULTRA / MEGA (plus private). Types: monthly, pay-per-use, volume tiers. Soft limits overage (“You **will** be charged”); hard limits return 429 ([consumer guide](https://docs.rapidapi.com/docs/consumer-quick-start-guide), [monetize tab](https://docs.rapidapi.com/docs/hub-listing-monetize-tab)). Platform free-plan cap on rapidapi.com: 1,000,000 requests/month. This harvest’s 11 calls were **33 credits**; signup grant is 100 credits; packs are on [/pricing](/pricing). We did not subscribe to a Hub listing, so there is no invented invoice here.
### Can one API replace RapidAPI’s Instagram, TikTok, and YouTube scrapers?
For the job in this post: yes, shown — four Author objects, three Post pages, a Prism creator-card, a URL lookup, and a 17-source search, one key, 33 credits. Official Graph / Research / Data API write-paths and user-OAuth flows are a different stack; this is public-data read. This does not claim every Hub listing on earth (weather, geo, payments) is in scope.
### Do I need a new API key per platform after leaving RapidAPI?
No. One `x-api-key`. RapidAPI also markets “a single API key” — the difference is N `X-RapidAPI-Host` values and N subscriptions versus one envelope. The four-row Author table above is the proof: YouTube, X, TikTok, Instagram, same field path.
### Is there a free RapidAPI alternative for social scraping?
Honest version: 100 welcome credits, a 0-credit catalog, then pay-per-call. That is not “the best free API platform.” Apify and RapidAPI BASIC $0 plans still meter objects and 429. Run the catalog curl before you spend anything.
If you are still comparing RapidAPI alternatives after the 11 calls: the catalog curl is free, [/pricing](/pricing) is public, and [when a managed API beats writing your own scraper](/blog/web-scraping) is the adjacent read.