The best Pytrends alternative in 2026 depends on the job, not on GitHub stars. pytrends-modern and other forks win a drop-in Python scrape that still talks to unofficial Google Trends HTML. SerpApi wins managed Google Trends JSON via engine=google_trends, with Free at 250 searches and Starter at $25 for 1,000. Trends MCP wins production agents that need Google Search plus other sources on MCP and REST. It is not a pytrends monkeypatch. GeneralMills/pytrends was archived on 17 April 2025 and is read-only. Google Search interest in pytrends sat at 7 on 22 August 2026 versus 68 a year earlier, down 89.71 percent, and versus 19 over 30 days, down 63.16 percent. The PyPI series for pytrends rose from 19.1 to 85.2 over 12 months, recent_volume 341,991 on 17 August 2026 (Trends MCP get_growth, pulled 26 August 2026). Brand search down is not package-download demand, and neither number is product quality.

Verdict first: three replacements, three different contracts

Job the Python project was hired forAlternativeWhy it wins that jobWhat it will not replace
Drop-in scrape of unofficial Google Trends HTML, familiar TrendReq callspytrends-modern and community forksActive PyPI, same mental model as the archived libraryA hosted SLA or multi-source index
Managed Google Trends HTML as JSON, without running SeleniumSerpApiengine=google_trends, documented fields, Free 250 / Starter $25 per 1,000TikTok, Amazon, Reddit, or npm in one schema
Production agents plus Google Search next to other demand sourcesTrends MCPMCP Bearer and REST, source google search, type Google TrendsA monkeypatch of from pytrends.request import TrendReq
Google-official consistently scaled relative interestGoogle Trends API Alpha1800-day window, ISO 3166-2 geo, announced 24 July 2025Self-serve access or an MCP tool until someone wraps it

Roundups that still rank pytrends-modern first for the overall query are answering a different question: “what do I pip install so old notebooks keep running.” That is a real job. It is not the job a SaaS cron, a Claude connector, or a Cursor agent hired. Treat those as separate purchases. The API-shaped sibling is best Google Trends APIs. The MCP-shaped sibling is best Google Trends MCP servers.

The 17 April 2025 archive closed the original library

GitHub marks GeneralMills/pytrends as archived by the owner on 17 April 2025, now read-only (github.com/GeneralMills/pytrends, accessed 26 August 2026). The project described itself as a pseudo-API. It wrapped the same unofficial web flows the browser used. Convenience was the product. Session cookies, 429s, and payload changes were the cost.

That archive date is the reason “just keep using pytrends” is no longer a production default. Old scripts still import it. New work that promises uptime to customers should assume the original package will not get a merge. A maintainer note on the repo in February 2025 already told readers to fork rather than wait for commit rights.

Google Search interest in the brand query pytrends sat at 7 on 22 August 2026 against 68 on the 12-month baseline (-89.71 percent) and against 19 on the 30-day baseline (-63.16 percent). Quote both index values. A quieter branded search term does not mean every fork stopped working. It means fewer people are googling the old package name.

The Python source on Trends MCP is a different series. Keyword pytrends, source python, 12-month window: 19.1 to 85.2 (+346.07 percent), recent_volume 341,991 dated 17 August 2026. That is PyPI-shaped demand, not Google Search. Installs can rise while the support-query cools. Neither series is a quality score for SerpApi or Trends MCP.

Forks still win the drop-in scrape

pytrends-modern is the closest replacement when the codebase already calls build_payload, interest_over_time, and related_queries, and the team wants to keep that object model. PyPI listed version 0.2.12 on 13 July 2026 (pypi.org/project/pytrends-modern, accessed 26 August 2026). The package page describes HTTP access, RSS, Camoufox browser mode, Selenium, async helpers, retries, and proxy rotation. Last-month downloads on that page were 30,890. Those are package facts, not an SLA.

Use a fork when the project is internal, Google Trends HTML is enough, existing tests expect DataFrames, and someone on the team can own 429s. Avoid it when a customer-facing product needs a contract that survives the next Google frontend change. A maintained unofficial wrapper is still unofficial. trendspyg and similar names show up in the same PyPI neighborhood. Check method coverage before a rewrite. This post does not invent a trendspyg version that was not opened in this run.

The original pytrends README is not an official Google Trends API. Forks that keep scraping the public UI inherit that limit. Google’s own Alpha, announced 24 July 2025, is a separate path: application-gated, consistently scaled relative interest, not absolute counts. See developers.google.com/search/blog/2025/07/trends-api (accessed 26 August 2026).

SerpApi is the pick when the application specifically needs Google Trends-style results as JSON and the team refuses to run browsers, cookies, or proxy pools. The engine string is google_trends. Documented data_type values include TIMESERIES (interest over time), GEO_MAP, GEO_MAP_0 (interest by region), RELATED_TOPICS, and RELATED_QUERIES (serpapi.com/google-trends-api, accessed 26 August 2026). Up to five queries per search apply to interest-over-time and compared geo breakdowns.

Public pricing the same day: Free $0 for 250 searches per month, Starter $25 for 1,000, Developer $75 for 5,000, Production $150 for 15,000 (serpapi.com/pricing). Cached searches on that vendor’s terms do not count. That grid fits prototypes and mid-volume SERP stacks that already pay SerpApi for Search or Maps. It does not include TikTok hashtags, Amazon best sellers, or npm downloads in the same response.

SerpApi is still a scrape of Google Trends surfaces, packaged as a product. The data model stays Google. If the next sprint needs Reddit Hot Posts or a get_growth window on YouTube, another provider appears. That is the honest limit, not a slight.

Trends MCP is the Pytrends alternative when the Python job is a production agent or a service that must ask Google Search and at least one other source without a second vendor. It does not monkeypatch TrendReq. There is no interest_over_time method to import. The contract is three tools: get_growth, get_trends (REST get_time_series), and get_top_trends.

MCP: https://api.trendsmcp.ai/mcp with Bearer. REST: POST https://api.trendsmcp.ai/api. Keyword series use source google search. The source string is not google trends. Live boards use type Google Trends. Default limit 25. Several percent_growth windows ride in one growth call. Free 100 requests per month. Starter $19 for 1,000. Pro $49 for 5,000. Business $199 for 25,000. Annual 20 percent off. Pricing (accessed 26 August 2026).

A notebook that only needs five Google keywords should not buy Trends MCP to look like an agent stack. A Cursor or Claude turn that must compare Google Search with TikTok or Reddit should not keep a local Selenium fork as the system of record. Those are different contracts. Trends MCP is first only for live multi-source trend data in agents. It is not the closest pip install to 2023 pytrends code.

Skip Trends MCP when legal requires Google-official Alpha series and access was granted. Skip SerpApi when the output must include Amazon or npm in the same schema. Skip the forks when the on-call rotation will not babysit HTML.

Sources

  • GeneralMills/pytrends archive banner: https://github.com/GeneralMills/pytrends (accessed 26 August 2026)
  • pytrends-modern 0.2.12 on PyPI: https://pypi.org/project/pytrends-modern/ (accessed 26 August 2026)
  • SerpApi Google Trends engine: https://serpapi.com/google-trends-api (accessed 26 August 2026)
  • SerpApi pricing: https://serpapi.com/pricing (accessed 26 August 2026)
  • Google Trends API Alpha announcement: https://developers.google.com/search/blog/2025/07/trends-api (accessed 26 August 2026)
  • Trends MCP get_growth on google search / pytrends and python / pytrends, pulled 26 August 2026

FAQ

Is pytrends-modern the best Pytrends alternative overall?

No. It is the best drop-in Python scrape for teams that already think in TrendReq. SerpApi is the better managed Google Trends JSON path. Trends MCP is the better production-agent path when Google Search is one source among several. Ranking the fork first for every “best Pytrends alternative” query hides those jobs.

Can Trends MCP replace from pytrends.request import TrendReq?

No. It does not expose the same methods, DataFrame shapes, or related-queries payload. Point the agent at MCP or REST. Rewrite the job as get_growth with source google search, or get_top_trends with type Google Trends. Keep a fork if the old unit tests must stay green.

How do SerpApi and Trends MCP compare on a 1,000-call month?

SerpApi Starter is $25 for 1,000 searches (serpapi.com/pricing, accessed 26 August 2026). Trends MCP Starter is $19 for 1,000 requests, with MCP and REST and sources beyond Google. The payloads differ. A SerpApi engine=google_trends search is a Google Trends page as JSON. A Trends MCP request is a growth window or a live board, including non-Google sources.

Did search interest in pytrends collapse because the library got worse?

Google Search moved from 68 to 7 over 12 months and from 19 to 7 over 30 days (Trends MCP pull 26 August 2026, recent date 22 August 2026). The PyPI series moved from 19.1 to 85.2 over 12 months, recent_volume 341,991 on 17 August 2026. Brand-query decline is not package-download demand. Neither series is a quality score. The operational fact that still matters is the 17 April 2025 archive.

Is the official Google Trends API a Pytrends replacement in Python today?

Only for accepted Alpha testers. Google announced it on 24 July 2025. Access is application-gated. The series is consistently scaled relative interest, not absolute counts, over a rolling 1800-day window. There is no public self-serve price on developers.google.com (accessed 26 August 2026). Everyone else still chooses a fork, SerpApi, or a multi-source API.