This article tracks the release age and training cutoff for 20 current AI models across 8 labs, providing a staleness metric that counts upward from each model's live release date. The training cutoff is the date a model stopped reading, and it can be manually checked by asking the model directly. The data is updated live and can be accessed as a JSON file. AI summary
Firehose
Filtered to Hacker News, tagged “machine learning maintenance” · clear filters
Browse: People · Companies · Papers · Podcasts · Hacker News · Deep dives