Unsloth Studio is a web UI for training and running open models like Gemma 4, Qwen3.6, DeepSeek, gpt-oss locally.
unslothai/unsloth is accelerating — star growth is speeding up by +13.6/day (96th percentile in the tracked cohort), on top of +58.4 stars/day. A recent release is helping fuel the run.
Reconstructed from 352 snapshots — the time-series GitHub’s API doesn’t expose.
Release count rose to 69.
Release count rose to 68.
Release count rose to 67.
Live score, refreshed every 6 hours. Links back to this page.
[](https://breakwave.vercel.app/repo/unslothai/unsloth)Elevated breakout odds over the next 14 days, led by star acceleration (96th pctl) and star velocity (93th pctl). Confidence is high given the available history.
Transparent heuristic · logged for model training
Percentile rank within the tracked cohort. The score self-calibrates — it answers “accelerating vs. everything else,” not raw size.
Release count rose to 66.
~24 forks/day, 3.3× the trailing baseline.
Release count rose to 65.
~28 forks/day, 4.5× the trailing baseline.
~14 forks/day, 4.0× the trailing baseline.
Release count rose to 64.
~14 forks/day, 5.0× the trailing baseline.
~10 forks/day, 4.4× the trailing baseline.
Release count rose to 63.