What this is

New Term Radar watches the places where AI vocabulary is born and reports new terms the day they start to move — before they show up in search trends or industry news.

Signals

  • GitHub breakout — repositories created in the last two weeks whose star velocity is far above the pack. A brand-new repo pulling hundreds of stars a day usually carries a new name with it.
  • Hugging Face trending — the daily most-downloaded model list. We diff it against yesterday, so a model that just entered the top 100 is a fresh name.
  • New pages on AI sites — sitemaps of AI tool directories and model hubs, diffed day over day; new URLs are new products, new model names, new vocabulary.

How a term gets on the radar

Each signal produces candidates. A candidate becomes a tracked term when it looks like a name rather than a sentence, and when it clears that signal's momentum threshold. No editorial pick: if it moved, it is listed, and the evidence is printed next to it.

How to read it

A radar hit means early, not important. Most new names fade. The value is in seeing them days or weeks before they are obvious, with the raw numbers attached so you can judge for yourself.

Freshness

The scan runs daily and every run is archived — see the archive. Nothing is edited after publication.

Built on

This site renders the output of the open-source new-term-radar pipeline (MIT-licensed), plus a daily diff of the Hugging Face top-100. Questions: support@jev001.org.