Benchmarking Frontier AI models on Advanced Mathematic and Visual Programming -- A challenging benchmark where the best models still perform poorly

Fable on shader bench

Only partially run, but Mythos appears to be absolutely mogging this benhcmark