100 under-the-radar AI repositories: the strongest picks sit between 500 and 30,000 stars

Asked:

“What are the best community-recognized GitHub repositories for AI enthusiasts? I'm interested in mid-tier, lesser-known projects rather than the most popular ones. Please give me the coolest ones available right now, at least a top 100.”

A live-web scan on 2026-10-21 surfaced 117 distinct AI-related GitHub repositories recommended by curated roundups, technical comparisons, community collections and research-resource pages across at least five hosts; this report ranks the top 100 of those 117 by a discovery score that rewards curated-source mentions and the 500–30,000-star band. At the row-recommendation level the research is single-sourced: 51 of the 100 rows trace to one large curated list, while repository identity and star counts come independently from each GitHub page.

The discovery landscape: stars vs discovery score

Each dot is one repository (81 of 100 shown; 18 lack an extractable star count and one has 0 stars). Stars on a log scale. Hover a dot for details; click a legend entry to toggle a band.

Where the 100 sit by recognition band

Count of repositories per band; label shows average stars where available.

microsoft/semantic-kernel leads at #1 with 28,656 stars and 3 independent curated-source mentions — exactly the recognized-but-not-famous profile the score rewards. dev.to

statewright/statewright reaches #10 with only 506 stars because two separate comparisons recommend it — the smallest project in the top ten. dev.to

The 18 mainstream-spillover repositories above 50,000 stars average 137,458 stars, nearly 8× the mid-tier band's 17,628 — they are kept, de-emphasized, rather than dropped. dev.to

Eleven micro projects under 500 stars (average 142) make the list, the best of them at #57 — mostly research-adjacent collections such as awesome world-model and multimodal-prediction lists. arxiv.org

The full top 100

#	Repository	Why it is cool	Category	Stars	Mentions	Score	Source

Method: 100 of 117 AI-related GitHub repositories found in a live-web scan on 2026-10-21, one row per repository, ranked by an editorial discovery score that rewards independent curated-source mentions and the 500–30,000-star range, gives smaller bonuses to emerging/micro and 30,000–50,000-star projects, and penalizes repositories above 50,000 stars, with a logarithmic star term; it is a discovery heuristic, not a quality benchmark. Star counts were fetched from repository pages where available; 18 repositories have no extractable star count and are left blank, never zero. Categories reflect source wording and may overlap. 51 of 100 recommendations trace to a single curated list (single-sourced at the row level); 17 of the 117 candidates were cut for space.

This report was generated automatically by Keenable SELECT at a user's request, from publicly available web sources linked herein. Keenable does not review, verify, or endorse its contents and makes no representation as to accuracy, completeness, or timeliness; AI-based extraction may contain errors. Nothing in this report is investment, legal, financial, or other professional advice. All trademarks and referenced content remain the property of their respective owners; no affiliation or endorsement is implied. To report an error, rights concern, or request removal: legal@keenable.ai.