Singularity Gate benchmark: Fable 5 and GPT-5.6 lead

The Singularity Gate benchmark evaluates AI models on predicting paradigm-breaking scientific discoveries after their training cutoff. Fable 5 and GPT-5.6 lead the ranking, outperforming other models in this novel test. The benchmark tests whether AI can extrapolate beyond its training data to identify truly novel breakthroughs.