Covering the AI beat, every day.
Model Beat is an AI news tracker and model comparison site. It watches about 45 feeds around the clock, collapses the fifteen articles covering the same event into one story, ranks what is left by significance, and links the news to the models it is about. The model tracker follows 140+ models with benchmark scores, pricing, and specs, refreshed every few hours.
How it works
A pipeline runs every 3 hours. It ingests from roughly 45 configured feeds (last month the stories it collected came from more than 1,300 different outlets), groups articles that cover the same event, and scores each story on source authority, breadth of coverage, and impact. Nothing is published without a citation: every story shows its original sources with direct links and timestamps, and the site never reproduces full article text.
Where the data comes from
Benchmark scores come from Epoch AI (CC BY) and Artificial Analysis. Pricing, context windows, and model specs come from OpenRouter. News comes from the publishers credited on every story. When sources disagree or a model has no independent scores yet, the page says so instead of filling the gap.
Who makes it
I’m Anmol Gaur, a data engineer. I built Model Beat solo because I kept finding out about model releases from three-day-old tweets, and I run it as an independent project. The code, the pipeline, and the editorial calls are mine. You can find me on LinkedIn, X, and GitHub, or get the weekly digest at themodelbeat.com/digest.
Corrections and contact
If something here is wrong, tell me and I will fix it: email anmolgaur45@gmail.com or reply to any digest issue. Corrections get made in place, quickly.