Move over Claude, 99.9% AR, 77.2% Beam – No RAG, No Embeddings, No Tricks

CEM888AI released benchmark results showing 99.9% accuracy on AR and 77.2% on Beam without using RAG, embeddings, or other common techniques, outperforming models like Claude.

CEM888AI / CEM888.AI-Site Public Notifications You must be signed in to change notification settings Fork 0 Star 0 Expand file tree / benchmarks / Copy path Directory actions More options More options Directory actions More options More options Latest commit History History History / benchmarks / Copy path Folders and files Name Name Last commit message Last commit date parent directory .. AR-Results-99.9pct.md AR-Results-99.9pct.md Vetta-BEAM-Honest-77.2pct.md Vetta-BEAM-Honest-77.2pct.md beam-full-results.html beam-full-results.html beam question contexts.json beam question contexts.json vetta beam v9 results.jsonl vetta beam v9 results.jsonl vetta live results.jsonl vetta live results.jsonl