Launched this week
OpenSourceAI.tech
Discover 10,000+ open-source AI projects, models & tools
14 followers
Discover 10,000+ open-source AI projects, models & tools
14 followers
OpenSourceAI is the discovery hub for the open-source AI ecosystem. Explore 10,000+ AI projects, models, tools and frameworks in one place. Discover trending repositories, compare AI models, track new releases, find open-source alternatives and explore the latest innovations in AI. Whether you're a developer looking for your next project, a researcher exploring new models, or a builder searching for open-source alternatives, OpenSourceAI helps you find what matters.
The curation angle here is great, but it would help a lot if you added a "fork relationships" graph so I can see how projects branch off each other and trace lineage between models like Llama derivatives. Right now everything feels flat when I try to compare which repos actually share DNA.
@atakansiler Fork lineage and model lineage are actually two different graphs — and you're pointing at the more interesting one. GitHub forks tell you about code (SD WebUI Forge descends from AUTOMATIC1111), but Llama derivatives aren't forks: they're retrained weights, and that lineage lives in Hugging Face base_model metadata.
We're going to measure coverage there first, then surface it on each model page — "derived from X, N known descendants", with links both ways — before deciding whether a visual graph earns its place. Thanks for the push, this one's going on the roadmap.
@atakansiler Thanks for this — it sent us down a rabbit hole, and you were right that everything felt flat. We shipped the feature this week.
One honest finding first: we tested the literal "fork relationships" graph and killed it. We sampled our 10,000 repos against the GitHub API — zero declare a parent. Curated directories select canonical repos, and serious projects start fresh instead of clicking Fork (even SD WebUI Forge isn't a GitHub fork of AUTOMATIC1111). That graph would have been empty.
So we measured DNA differently: shared GitHub topics, weighted by rarity. "ai" (2,000+ repos) proves nothing; "ggml" (12 repos) means you're the same technical family. That gives us 1,101 families covering ~8,000 projects:
• The map: https://opensourceai.tech/dna.html
• A tight family: https://opensourceai.tech/dna/ggml.html — llama.cpp with rwkv.cpp, bark.cpp, clip.cpp orbiting it
• A big one: https://opensourceai.tech/dna/comfyui.html
• On every project page, a "Shares DNA with" block now shows the closest relatives with the exact topic they share — no opaque score, just verifiable facts: https://opensourceai.tech/project/ggml-org-llama-cpp.html
For model lineage (Llama derivatives etc.), that's a different graph — retrained weights, not forked code. We pull it from Hugging Face base_model declarations and show it on model pages where authors declared it. We only display exact matches; guessed ancestry doesn't belong in a directory.
Would love your take on whether the constellations read well — this shipped because of your comment.
The collection looks great already, but adding a simple "Weekly Digest" email with the top trending repos and releases would be super helpful for staying in the loop without having to check the site daily.
@sudenurdolaner Thanks — that's exactly the gap. Right now the digest only goes to signed-in members who follow specific projects, so there's no way to just stay in the loop passively.
We're shipping a weekly digest you can subscribe to with just an email, no account needed: top trending repos of the week, new releases, and the models that moved on price or benchmarks. Signup form goes live within 24h, first issue next week.
Appreciate the nudge — this one jumped the queue because of your comment.
Finally a clean way to browse the chaos of open source AI stuff. The trending repos section actually surfaced a few gems I hadn't come across yet.
@lyas1059603 Thank you! That was exactly the goal—to cut through the noise and make discovering great open-source AI projects much easier. I'm really happy the Trending Repositories section helped you find new gems. I'll keep improving it with better ranking signals and more insights as the ecosystem grows. Thanks for the feedback! 🙌
the model comparison view is genuinely useful, saved me a bunch of time digging through github repos
@kamile1170180 Thanks! I'm really glad you found it useful. Saving developers time is one of the main goals of OpenSourceAI. More comparison features are on the roadmap! 🚀
Finally a clean way to browse open-source AI without drowning in GitHub search results. The trending section helped me find a smaller model I hadn't seen before, which was a nice surprise.
@hayrettin960311 Thank you! That was exactly the goal—to make discovering open-source AI simple and help great projects get the visibility they deserve, not just the biggest repositories. I'm really glad the Trending section helped you uncover a model you hadn't seen before. Thanks for the feedback, and there's plenty more to come!
finally a clean way to browse the open-source AI scene without drowning in github search results. loved how quickly i could compare models side by side
@azatv0rh Thank you! That was exactly the idea behind OpenSourceAI: make discovering open-source AI fast and enjoyable instead of digging through endless GitHub search results. I'm really glad the side-by-side model comparison was useful—that feature will continue to improve with more metrics and insights. Thanks for the feedback! 🙌