Stars per year: comparing old and new GitHub repos

Total stars favor old repos, because they have had longer to collect them. Divide a repo’s stars by its age and the picture changes: of the 88 repos in the top 100 by total stars that are at least a year old, only 44 are also in the top 100 by stars per year.

All counts are for the 2,500 repos that have a page on RepoTrending on Oct 23, 2026, the most starred of the 12,468 we track. This guide is for a developer who sees a 2025 repo and a 2014 repo with similar star counts and wants to know which one gathered attention faster. The stars-per-year figure is our own arithmetic on one snapshot; the site doesn’t rank by it. It is a fair question and a weak answer, and we say where it breaks.

How do you compare an old repo with a new one?

Divide the stars by the age in years. We took the days from each repo’s creation date to Oct 23, 2026, divided by 365.25, and divided the star count by that number.

The top five by total stars are all old repos, created between 2013 and 2018:

The top five by stars per year, among repos at least a year old, are all from 2025:

The lowest of the second five, anomalyco/opencode at 148,700, is more than double the highest of the first, codecrafters-io/build-your-own-x at 65,700. You can open obra/superpowers or codecrafters-io/build-your-own-x to see the numbers we hold for each.

Does dividing by age change who is in the top 100?

Yes, mostly in favor of repos from 2024 or later. We compared two lists: the top 100 by total stars, and the top 100 by stars per year among the 2,296 repos at least a year old. 12 of the top 100 by total stars are under a year old and can’t be in the second list, so the fair overlap is 44 of the other 88.

Read the table as a change of lens, not a verdict on which repos matter. The five old repos in the first table all stay in the top 100 by stars per year, but far from the top: codecrafters-io/build-your-own-x ranks 22nd, public-apis/public-apis 46th, sindresorhus/awesome 52nd, freeCodeCamp/freeCodeCamp 61st and EbookFoundation/free-programming-books 90th. A lower rank on this measure doesn’t mean a repo is used less; it collected its stars over more years.

What does the age divide hide?

It hides how young a repo is, and a small age makes the rate explode. That is why we left out the 204 repos under a year old. morluto/rea, created Apr 14, 2026, has 366,708 stars in about 0.53 years, which comes to roughly 697,600 stars per year. storytold/photocraft, created Sep 30, 2026, has 71,410 stars after 23 days, which works out to about 1,134,000 per year. Nobody should assume either pace holds for a year.

Group the repos by the year they were created and the pattern shows:

The median total hardly moves, from 28,530 to 33,612, while the median per year goes from 2,100 to 62,900. The first is not a finding about GitHub. Every repo on our list has at least 18,279 stars, so an old repo that stayed small is not here, and a young one needs a big start to get in. The second is mostly the division: the same 30,000 stars over 0.5 years or over 12 years.

Why not rank by stars per year?

Because a measure that rewards youth has the same flaw as one that rewards age, only pointing the other way. The age comes from the date the GitHub repo was created, which can differ from the date the project started. A repo that moved to GitHub late looks younger than it is.

What the site ranks by is the change in stars over a window, counted from daily snapshots. Stars gained vs total stars compares the two measures, and daily, weekly or monthly explains the windows. Our first daily gains arrive Oct 24, the first weekly ones Oct 30 and the first monthly ones Nov 22. When they do, the figures here can be checked against real gains. For the other things to look at before you depend on a repo, read how to check a GitHub repo before you depend on it.

What does this sample hide?

It hides every repo outside the 2,500 and every moment between snapshots. We take every public repo with 5,000 or more stars plus recent ones above a star floor, and only the 2,500 most starred get a page, so the old repos that never grew are missing. How we rank lists the full method and what 12,468 GitHub repos look like counts the list by language, age and topic.

We don’t check for bought or farmed stars yet, so every star count here is what GitHub displays and nothing more. The list also changes from Oct 24, when the biggest gainers start to come first, so these numbers will move.

Frequently asked questions

How do you compare a new GitHub repo with an old one?

Divide its stars by its age in years. On Oct 23, 2026, the five repos with the most stars per year, among those at least a year old, were all created in 2025, while the five with the most total stars were created between 2013 and 2018.

Are stars per year a fair measure?

Only a rough one. It assumes a repo collected its stars at a steady pace, and a repo that got most of them in one week looks the same as a steady one. The measure that shows pace is stars gained over a window, which our rankings start to show on Oct 24.

Why do you leave out repos under a year old?

A few weeks of age makes the number explode. storytold/photocraft was created Sep 30, 2026 and has 71,410 stars, which works out to about 1,134,000 stars per year, a pace nobody can assume will last.

Do newer repos have fewer stars than older ones?

Not on our list. Among the 2,500 repos with a page, the median total is about the same for every creation year, between 28,530 and 33,612, because every repo on the list has at least 18,279 stars.

Where to look next

• The all-time list: the most starred repos, each with a page you can open to see its stars and creation date.
• AI agent repos: the 165 repos under ai-agents, mcp and coding-agents.
• How we rank: how the 12,468 repos are found and how star gains will be counted.

Data used

Source for every row: RepoTrending, GitHub REST API (public repository search, read-only), snapshot of Oct 23, 2026 (collector export repos.json, run of 2026-10-23), counted by RepoTrending. Stars per year = stars on Oct 23 divided by ((Oct 23, 2026 minus the creation date, in days) divided by 365.25), rounded to the nearest 100. Medians of an even count are the mean of the two middle values.

More guides

• How to check a GitHub repo before you depend on it
• Daily, weekly or monthly: which window to watch?
• Find Claude Code repos worth trying (Oct 2026)
• GitHub trending alternatives: what each one counts
• How does GitHub decide what is trending?
• Open source licenses on popular GitHub repos
• Stars gained vs total stars: which says more?
• Trending badge for your README, and what it says
• What 12,468 GitHub repos look like