A high-throughput and memory-efficient inference and serving engine for LLMs
vllm-project/vllm is drawing steady momentum at +87.3 stars/day (95th percentile in the tracked cohort), for a momentum score of 89.4/100. 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 106.
Release count rose to 105.
Release count rose to 104.
Live score, refreshed every 6 hours. Links back to this page.
[](https://breakwave.vercel.app/repo/vllm-project/vllm)Elevated breakout odds over the next 14 days, led by star acceleration (88th pctl) and star velocity (95th 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 103.
Release count rose to 102.
Release count rose to 101.
~144 forks/day, 3.2× the trailing baseline.
Release count rose to 100.
Release count rose to 99.
~68 forks/day, 3.7× the trailing baseline.
~91 forks/day, 4.8× the trailing baseline.