Astra vs Fable 5.1: The Wildest Community Demos From the New Model Race

eyishazyer · x · 2026-09-08

eyishazyer rounds up the wildest things people are building with GPT-6 Astra and Fable 5.1, arguing the model race just got a lot more interesting.

The thread also includes a notable technical take: Fable 5.1 is weak at Blender tasks because it hasn't seen enough Blender training data — much like older models were bad at chess until trained on it — suggesting such gaps can likely be closed with more data.

Original post →

More from Models

Models channel →