Trends API - official Python client

One score. Every platform. Official Python client for the Trends API - Google, TikTok, Amazon, Reddit, YouTube, Steam, npm and 30+ sources normalized to a single 0-100 trend score.

Get a free API key → · Quickstart → · llms.txt →

Why Trends API (not scrapers, not a SERP swiss-army knife)

Also: failed requests do not burn quota - only successful 200s count. Plan limits are predictable; you are not fighting upstream rate limits to get a single series.

Replacing archived pytrends? Same job (Google interest over time), without the breakage - and you also get TikTok, Amazon, Reddit, YouTube, live feeds, and growth windows.

Install

pip install trendsapi

Python 3.9+. Only dependency: httpx.

Quickstart

Cross-platform compare in a few lines:

from trendsapi import TrendsAPI

client = TrendsAPI(api_key="YOUR_API_KEY")  # or export TRENDSAPI_KEY=...

keyword = "creatine gummies"

# 1) Growth on three platforms - same keyword, comparable scores
for source in ("google search", "tiktok", "amazon"):
    g = client.get_growth(source=source, keyword=keyword, percent_growth=["12M"])
    r = g.results[0]
    print(source, r.growth, r.direction)

# 2) ~5 years of weekly history
series = client.get_time_series(source="google search", keyword=keyword)
print("latest", series[-1].date, series[-1].value)

# 3) What’s trending right now (no keyword)
hot = client.get_top_trends(type="TikTok Trending Hashtags", limit=5)
print(hot.data)

Async (fan out sources)

import asyncio
from trendsapi import AsyncTrendsAPI

async def main():
    client = AsyncTrendsAPI()  # reads TRENDSAPI_KEY
    google, tiktok, amazon = await asyncio.gather(
        client.get_time_series(source="google search", keyword="air fryer"),
        client.get_time_series(source="tiktok", keyword="air fryer"),
        client.get_time_series(source="amazon", keyword="air fryer"),
    )
    print(len(google), len(tiktok), len(amazon))

asyncio.run(main())

Auth & quota

1. Free key in ~60 seconds: trendsapi.ai/#get-key (100 requests/month).
2. Pass TrendsAPI(api_key=...) or set TRENDSAPI_KEY.
3. Only successful 200 responses count. 4xx/5xx are free.
4. Plan tiers scale when you need more volume - see pricing.

This is the opposite of scraper life: you don’t burn hours on IP bans; you burn a clear monthly lookup budget.

Three modes

Full source and feed lists: trendsapi.ai/llms.txt

Keyword sources

google search · google images · google news · google shopping · youtube · tiktok · reddit · amazon · wikipedia · news volume · news sentiment · app downloads · app rankings · npm · steam

from trendsapi import KEYWORD_SOURCES, LIVE_FEEDS

MCP (Claude, Cursor, VS Code, ChatGPT)

{
  "mcpServers": {
    "trendsapi": {
      "url": "https://api.trendsapi.ai/mcp",
      "transport": "http",
      "headers": { "Authorization": "Bearer YOUR_API_KEY" }
    }
  }
}

Then ask: “Compare 12-month growth of creatine gummies on Google, TikTok, and Amazon.”

Errors

from trendsapi import TrendsAPI, TrendsAPIError

try:
    TrendsAPI(api_key="bad").get_growth(source="google search", keyword="x")
except TrendsAPIError as e:
    print(e.status, e.code, e.message)

If a growth preset is longer than available history for that keyword (common on new TikTok tags), the API may still return HTTP 200 with a per-row status="error". Those rows parse with growth=None and error / message set.

Links

• Product: https://trendsapi.ai
• Quickstart: https://trendsapi.ai/#quickstart
• API for agents: https://trendsapi.ai/llms.txt
• All sources: https://trendsapi.ai/trends
• GitHub: https://github.com/trendsapi/trendsapi-py

License

MIT. Data access follows your Trends API plan limits.