pytrends-ng: pytrends without the 429 errors

pytrends was archived in April 2025 and its last release is from 2023. Google answers it with 429 Too Many Requests most of the time. pytrends-ng keeps the same calls, so your code works again after changing one import:

# from pytrends.request import TrendReq
from pytrends_ng import TrendReq

pytrends = TrendReq()  # reads APIFY_TOKEN from the environment
pytrends.build_payload(["chatgpt", "claude", "gemini"], timeframe="today 3-m", geo="US")

pytrends.interest_over_time()        # DataFrame: date index, one column per keyword, isPartial
pytrends.interest_by_region()        # DataFrame: one row per region
pytrends.related_queries()           # {keyword: {"top": DataFrame, "rising": DataFrame}}
pytrends.trending_searches(pn="united_states")

How it works

The requests run on a hosted Google Trends actor on Apify, which rotates IPs and retries rate limits for you. No proxies to buy, no sleep loops.

1. Create a free Apify account. It includes $5 of usage every month.
2. Copy your API token from Settings > Integrations.
3. pip install pytrends-ng and export APIFY_TOKEN=...

Pricing is $2 per 1,000 results, where one result is one keyword in one country with its full timeline, regions and related queries. A 5-keyword payload costs $0.01, so the free credit covers about 2,500 keywords a month.

Better than the original

• More than 5 keywords. build_payload takes any number of keywords and puts them all on one shared 0-100 scale. Google's own limit is 5.
• Small keywords stay accurate. Keywords far smaller than the biggest one keep decimal values instead of rounding to 0.
• One request per payload. interest_over_time, interest_by_region and related_queries all read from the same run.

Supported calls

Constructor arguments such as hl, tz, proxies and retries are accepted and ignored, so existing code does not break.

License

MIT. Not affiliated with Google or the original pytrends authors.