Fable 5.1 benchmarks show insane jumps on Terminal/Science

kimmonismus · x · 2026-09-02

Benchmark results for Fable 5.1 are insane, showing unexpected massive jumps on Terminal-Bench 4.0, Science-Bench, and HLE. This sets the stage for OpenAI and Astra to respond.

Original post →

More from Models

Models channel →