Fable 5.2 Tops MazeBench Token Efficiency but May Have Memorized Answers
Fable 5.2 leads MazeBench in token efficiency, scoring 21% at the 200M-token checkpoint, but the benchmark author found it answering with an outdated solution from a puzzle updated two weeks earlier, suggesting possible memorization.
2026-11-16 ~ 2026-11-16 · 2 related posts
- Fable 5.2's Puzzle Skills Look Suspiciously Good, Reciting Outdated MazeBench Solution — patience_cave · 2026-11-16
- Fable 5.2 Leads MazeBench Token Efficiency; Billion-Token Runs Underway — patience_cave · 2026-11-16