YouTube Data API v3 Guide

YouTube Data API v3 Guide | Generated by AI

Home 2026.07

Question: How to use YouTube Data API v3?

Answer:

YouTube Data API v3 is the official REST API for interacting with YouTube programmatically. You can use it to:

Search videos/channels
Get video metadata
Upload videos
Manage playlists
Read comments
Manage your own channel

For CLI automation (your previous question about uploading videos), the important part is OAuth2 + videos.insert.

Official docs: YouTube Data API v3 Documentation

1. Create Google Cloud project

Go to:

Google Cloud Console

Create project:

my-youtube-cli

Enable:

APIs & Services
    ↓
Library
    ↓
YouTube Data API v3
    ↓
Enable

2. Create OAuth credentials

For uploading videos, API key is not enough.

Need:

APIs & Services
    ↓
Credentials
    ↓
Create Credentials
    ↓
OAuth client ID

Choose:

Desktop application

Download:

client_secret.json

Example:

youtube-cli/
 ├── client_secret.json
 └── upload.py

3. Install Python SDK

pip install google-api-python-client google-auth-oauthlib

4. Upload video example

Minimal uploader:

from googleapiclient.discovery import build
from googleapiclient.http import MediaFileUpload
from google_auth_oauthlib.flow import InstalledAppFlow


SCOPES = [
    "https://www.googleapis.com/auth/youtube.upload"
]


flow = InstalledAppFlow.from_client_secrets_file(
    "client_secret.json",
    SCOPES
)

credentials = flow.run_local_server(
    port=8080
)


youtube = build(
    "youtube",
    "v3",
    credentials=credentials
)


request = youtube.videos().insert(
    part="snippet,status",
    body={
        "snippet": {
            "title": "My AI generated video",
            "description": "Uploaded by CLI",
            "tags": [
                "AI",
                "LLM"
            ],
            "categoryId": "28"
        },
        "status": {
            "privacyStatus": "private"
        }
    },
    media_body=MediaFileUpload(
        "video.mp4",
        chunksize=-1,
        resumable=True
    )
)


response = request.execute()

print(response["id"])

Run:

python upload.py

First time:

Open browser
↓
Login Google account
↓
Allow YouTube access
↓
Token cached

After that, your CLI can upload automatically.

5. Search videos

Example:

youtube.search().list(
    part="snippet",
    q="GPT-5 tutorial",
    maxResults=10,
    type="video"
).execute()

Response:

{
 "items": [
   {
     "id": {
       "videoId": "abc123"
     },
     "snippet": {
       "title": "GPT-5 tutorial"
     }
   }
 ]
}

6. Get video information

youtube.videos().list(
    part="snippet,statistics",
    id="VIDEO_ID"
).execute()

Returns:

{
 "statistics": {
   "viewCount": "12345",
   "likeCount": "500"
 }
}

7. Build a real AI video pipeline

For an AI content factory, architecture could be:

                LLM Agent
                    |
                    v
          Generate Script
                    |
                    v
        TTS / Avatar / Video Model
                    |
                    v
              ffmpeg
                    |
                    v
             video.mp4
                    |
                    v
        YouTube Data API v3
                    |
                    v
              Published Video

CLI:

youtube upload \
    --file video.mp4 \
    --title "AI News #001" \
    --privacy public

Internally:

ww
 |
 +-- youtube.py
 |
 +-- oauth.py
 |
 +-- upload.py

8. Quota limits

YouTube API has quota:

Default:

10,000 units/day

Important costs:

Operation	Cost
search.list	100 units
videos.list	1 unit
upload video	1600 units

So:

10000 / 1600 ≈ 6 uploads/day

Default quota is enough for personal automation, but a video factory may need quota increase.

9. For multiple channels

Each channel needs:

Google account
      |
OAuth authorization
      |
refresh_token
      |
YouTube API

Store:

accounts/
 ├── channel_A.json
 ├── channel_B.json
 └── channel_C.json

Then:

youtube = build(
    "youtube",
    "v3",
    credentials=channel_A_credentials
)

For your AI-agent workflow, YouTube Data API is actually a good fit: an agent can generate content, call ffmpeg, upload, check analytics, and iterate. The main bottleneck is usually content generation + thumbnail + distribution, not the API itself.

Back Donate