Hugging Face Search rose while huggingface fell

Store hugging face and huggingface as two keywords. On 2026-09-05, Google Search for hugging face scored 41.0 for the week of 2026-08-29, up 28.12% from 32.0 on 2026-08-22. The one-word token huggingface scored 31.0, down 13.89% from 36.0. Keyword News for hugging face was already 93.0, up 27.4% week over week. Google News Top News ranked a Forbes row on Nvidia acquiring Hugging Face at 15 as of 2026-09-05T08:01:32Z. That live board is not last week's Search close. YouTube for hugging face sat at 53.0. The spaced token and the concatenated token moved in opposite directions on the same 7D window. Do not join them.

How far apart are hugging face and huggingface on Search?

Pulled on 2026-09-05 with source set to google search. Recent points land on 2026-08-29.

A 7D alert on hugging face fires as an increase. The same window on huggingface fires as a decrease. Scores are 0-100 per keyword. 41.0 is not "larger attention" than 31.0 across tokens. Keep two rows.

get_time_series for hugging face on Google Search puts the 41.0 tip in a 261-point weekly pull. The series peak is 100 on 2026-07-25. That week Search sat at 100 while News sat at 62. Later weeks: 64 on 2026-08-01, 45 on 2026-08-08, 42 on 2026-08-15, 32 on 2026-08-22, then 41 on 2026-08-29. The 7D bounce is a rebound off 32, not a new high. The 3M drop of 54.95% is the leftover from that July peak.

Nvidia Search on the same day scored 45.0 for the week of 2026-08-29, up 12.5% from 40.0, with estimated volume 4,760,000. That 45.0 is a different keyword. Do not add it to 41.0.

POST https://api.trendsapi.ai/api
Authorization: Bearer <api_key>
Content-Type: application/json

{"mode":"get_growth","source":"google search","keyword":"hugging face","percent_growth":["7D","30D","3M","12M"]}

The response envelope is {"statusCode": int, "body": string}. Parse body a second time before reading results. A first-pass decode that treats body as an object misses recent_value. The same POST with keyword set to huggingface is a second request. Caps: free 100 successful 200s per month, then 5,000 / 25,000 / 100,000. Two tokens times Search, News, and YouTube is 6 calls before any live board.

Source field notes sit on Google Trends. A C# helper that stores those rows is C# Trends API.

Did keyword News and Google News Top News agree?

They did not. Keyword News is a weekly series on the exact string. Google News Top News is a live ranked board with no keyword.

Keyword News for hugging face was 0 from the start of the 2021-09-11 pull through 2026-02-07. First non-zero week is 4 on 2026-02-14. Then 8, 17, 14, and a spring band in the teens. 2026-07-11 is 0 again. 2026-07-25 jumps to 62 (the Search 100 week). After that, News kept climbing while Search fell: 42, 33, 58, 73, then 93 on 2026-08-29. News is one week of 93.0 at the tip. Search is 41.0 on the same date. A 12M News growth of 999999 against a 0.0 baseline is the low-quality label. Print 93.0 versus 0.0.

huggingface News is a thinner series. 33.0 on 2026-08-29 versus 0.0 on 2026-08-22 also returns 999999. Print 33.0. The 3M compare against 34.0 is -2.94%. Do not treat 93.0 and 33.0 as one coverage number.

Google News Top News top 15 as of 2026-09-05T08:01:32Z put the Forbes Nvidia Hugging Face row at rank 15. That board is a headline list. It is not the 93.0 keyword-News tip. The weekly series still ends on 2026-08-29. The live board is dated 2026-09-05. Store as_of_ts on board rows and recent_date on growth rows. Do not write rank 15 onto score 93.0.

Nvidia keyword News scored 58.0 on 2026-08-29, down 42.0% from 100.0 on 2026-08-22, up 205.26% from 19.0 on 2026-08-01. The company token peaked a week before the Hugging Face News tip. Pair them. Do not replace hugging face with nvidia because a headline contains both words.

What did YouTube, Wikipedia, Reddit, and PyPI return?

Same pull day. YouTube uses the 2026-08-29 week. Wikipedia is monthly (August 2026). PyPI (python) tips on 2026-08-24 and still lags Search by a week. Reddit resolved the request as huggingface with a tip dated 2026-09-05.

YouTube for hugging face is 53.0, up 29.27% over 7D and 194.44% over 3M. Search is down 54.95% over that same 3M. A YouTube climb does not cancel a Search drawdown. YouTube Trending top 15 as of 2026-09-05T10:01:40Z omitted Hugging Face. Keyword YouTube and the live YouTube board are different feeds.

Wikipedia keyword Hugging Face resolved as title Hugging face with 4,879 views and score 95.3 for 2026-08-01. July was 100.0 / 5,120 views (the peak in this pull). August 2025 was 11.4 / 586. Compare on recent_volume. Do not join 4,879 pageviews onto Search 41.0. Wikipedia fields are on Wikipedia Trends.

TikTok for hugging face returned 404 no_data. Reddit accepted the lookup and resolved it as huggingface. The Reddit tip dated 2026-09-05 scored 10.3 with volume 50, down 89.7% from 100.0 / 180 on 2026-08-29. That current week looks partial. Gate volume alerts on 2026-08-29 until the next complete Reddit week lands. PyPI has the same incomplete-tip pattern: transformers 47.0 / 23,753,091 on 2026-08-24 versus 100.0 / 50,014,124 on 2026-08-17. huggingface-hub 42.1 / 57,919,495 versus 90.4 / 124,048,150. Compare packages on recent_volume from the last complete week, not on the 0-100 tip. 124,048,150 downloads is not "twice" 50,014,124 in score space. The scores are 90.4 and 100.0.

Did the live Google Trends board show Hugging Face?

No. Google Trends top 25 as of 2026-09-05T08:01:32Z ran from what does a mistrial mean at rank 1 through lindsay clancy update at rank 25. Rank 9 was chatgpt. Rank 7 was tropical storm edouard. No Hugging Face token appears. 7d rank_change (baseline 2026-08-29T02:01:36Z) returned 7 movers; none were Hugging Face. 1d rank_change (baseline 2026-09-04T02:01:41Z) returned 25 movers; none were Hugging Face.

That is the split to store. Keyword News is 93.0. Search is 41.0 and still 59 points under the 2026-07-25 peak. The live Search board does not list the company. The live News board does, at rank 15, on a later timestamp than the weekly close. A job that only polls get_top_trends with type Google Trends misses the story. A job that only polls get_growth on huggingface sees a 13.89% 7D drop and misses the spaced-token bounce.

{"mode":"get_top_trends","type":"Google News Top News","limit":15}
{"mode":"get_growth","source":"google news","keyword":"hugging face","percent_growth":["7D","30D","12M"]}
{"mode":"get_growth","source":"google search","keyword":"hugging face","percent_growth":["7D","30D","12M"]}
{"mode":"get_growth","source":"google search","keyword":"huggingface","percent_growth":["7D","30D","12M"]}

Four calls. Parse body twice on each. Write as_of_ts from the board next to recent_date from growth. The C# client that posts those payloads is C# Trends API. Docs for the Search source are on Google Trends.