alt Hacker News> Frontier models score ~90% on Python but only 3.8% on esoteric languages, exposing how current code generation relies on training data memorization rather than genuine programming reasoning.
> Frontier models score ~90% on Python but only 3.8% on esoteric languages, exposing how current code generation relies on training data memorization rather than genuine programming reasoning.