YouTube API
10 endpoints read YouTube: videos, posts, channels, comments. From $0.0020 a call. No YouTube login and no cookie; a failed call is free.
One call, as it came back
A recorded call, fields already tidied up: ready for an AI or a spreadsheet.The first 2 rows of a response recorded 2026-10-12 18:50 UTC.
Run it in the Playground · $0.0020{
"code": "ok",
"request_id": "2f106ec5-c4a3-42e6-9af6-a2ffd5848816",
"data": {
"items": [
{
"id": "TMvQhQaH-rq",
"url": "https://www.youtube.com/watch?v=TMvQhQaH-rq",
"type": "video",
"title": "I tried meteor shower for 30 days. Here is what happened",
"author": {
"id": "UCee61G0GfCvYQ4K3j4BgnU5",
"username": "maya.makes",
"name": "Northside Coffee",
"avatar_url": "https://yt3.ggpht.com/slva_3aFP2TCzQu0m-ZM4MLuiZ8Oe90XwjxhzkFn=s88-c-k-c0x00ffffff-no-rj"
},
"metrics": { "views": 1840 },
"images": ["https://i.ytimg.com/vi/D0fStB6ZVGr/hqdefault.jpg"]
},
// … 19 more
]
},
"page": {
"next_cursor": "c1.8SROT5T8n_nqIVHBOaV7N-2SYvyHFY8uAkBYxJvF",
"cursor_expires_at": "2026-10-12T19:50:45.000Z",
"has_more": true,
"returned": 20
},
"warnings": [],
"meta": {
"capability": "youtube.videos.search",
"result_status": "complete",
"fetched_at": "2026-10-12T18:50:45.000Z",
"cached": false,
"replayed": false,
"unavailable": {
"text": "capability",
"lang": "capability",
"published_at": "capability",
"author.verified": "capability",
"author.followers": "capability",
"metrics.likes": "capability",
"metrics.comments": "capability",
"metrics.shares": "platform",
"metrics.favorites": "platform",
"metrics.reposts": "platform",
"hashtags": "capability",
"mentions": "capability",
"is_pinned": "platform",
"post_id": "platform",
"parent_id": "platform",
"root_id": "platform"
},
"billing": {
"amount": "0.002",
"currency": "USD",
"balance_after": "9.998"
}
}
}From a response recorded 2026-10-12 18:50 UTC. Names, text and IDs are examples, and so are the links: none is the recorded one (the page shows no one's personal data); the figures are as recorded. Long lists are cut. The request ID, cursor and balance are made up.
What you can get
Each row is an endpoint: what it returns and what one call costs.
What people build with it
Ready-made tasks your AI can run on YouTube data. Each is a few calls under a spending cap.
Comments and word of mouth
“What are people praising and complaining about under this post?”
Competitor account monitoring
“What did competitors post this week, and which post took off?”
Keyword search and mentions
“What is being said about this brand or keyword on each platform?”
Ask about YouTube right inside Claude or Cursor
Add the DataXAPI MCP server and the AI finds the YouTube endpoint it needs and calls it; answers link back to the source. You pay for the calls it makes.
https://api.dataxapi.com/mcpStart in three steps
The call shown here is YouTube video search. Every other YouTube endpoint is called the same way and answers in the same envelope.
Get a key
Sign in with an emailed code or Google. Free trial, no card needed.
Call it, or let your AI call it
One GET from your code. Or paste this prompt into your agent with your key in it: it sets itself up, then finds and calls these endpoints by itself.
Read the same shape every time
Items are in data.items (one object in data.item), the next page in page.next_cursor, the charge in meta.billing.amount. A failed call is not charged.
curl -G "https://api.dataxapi.com/v1/youtube/videos/search" \ --data-urlencode "query=meteor shower" \ -H "Authorization: Bearer $DATAXAPI_API_KEY"
from dataxapi import DataXAPI dx = DataXAPI() # reads DATAXAPI_API_KEY result = dx.youtube.videos.search(query="meteor shower") print(result["data"]["items"])
import { DataXAPI } from "@dataxapi/sdk";
const dx = new DataXAPI(); // reads DATAXAPI_API_KEY
const result = await dx.youtube.videos.search({ query: "meteor shower" });
console.log(result.data.items);dataxapi call youtube.videos.search 'query=meteor shower'{
"mcpServers": {
"dataxapi": {
"type": "http",
"url": "https://api.dataxapi.com/mcp",
"headers": {
"Authorization": "Bearer ${DATAXAPI_API_KEY}"
}
}
}
}Add this to your MCP client, then ask your agent to call youtube.videos.search.
Edge cases
What the YouTube endpoints do at their edges, and what to check when a call does not return what you expected.
- Page sizes
- A list returns 10 to 100 items a page, depending on the endpoint. Every list goes on to a next page:
page.next_cursor, passed back ascursor, each page a call of its own. - Fields it cannot give
- 9 of the 10 endpoints cannot fill every field of the object they return. Each endpoint's page lists which, and every answer repeats them in
meta.unavailable: a missing field is never a silent zero. - A missing object that reads as empty
- YouTube channel videos, YouTube comment replies, YouTube channel posts: an object that does not exist cannot be told from one with nothing under it, so the answer is an empty result, not a 404, and is charged as an empty result is.
- Channel ID or handle
- A channel is named by its ID (starting with UC) or its handle without the @. On the Posts tab, a handle nobody has is a 404, while an ID nobody has comes back empty and is charged as an empty result.
- Replies take a token
- Comment replies are asked for by
ext.replies_tokenfrom the video comments endpoint (full view only), not by the comment's ID; a comment ID is refused. - Shorts: one page
- With
type=short, a channel's Shorts come in one page with no next page, and without the channel on each. - Rounded figures
- Subscriber counts have three significant figures, as YouTube shows them; likes on comments and posts are rounded from a thousand up. Video search gives no likes, no description and no exact time published.
- No captions, no transcript
- A video without captions reads as not found and is not charged. Watch, shorts, embed, live and youtu.be links all work.
Questions about the YouTube API
Short answers. The details are on each endpoint's page.
How is the YouTube API priced?
Per call, no subscription. All 10 endpoints cost $0.0020 a call. You are charged when data comes back; a failed call costs nothing.
Do I need a YouTube account or a cookie?
No. One DataXAPI key is all you hold: no YouTube login, no cookie, no device to keep alive. The endpoints return publicly visible data only.
Is the data live?
A call fetches from YouTube at that moment, and meta.fetched_at says when. A result is kept for 24 hours, so a retry with the same Idempotency-Key is not charged again.
Is an empty result charged?
A list that comes back empty is charged its price: the search was run. Reading one object that does not exist answers not_found and is free.
What happens when YouTube changes something?
A call that fails is not charged, whatever the reason. An endpoint that is paused answers capability_paused, also free, and GET /v1/catalog/status says what is up right now.
Related
Call YouTube with one key.
Free trial on sign-up: up to 50 calls at $0.0020. No card. No YouTube login, no cookie, no subscription. A failed call is free.