kapynAI / Models

Benchmarks disagree on GPT-6 Astra, but its human-beating efficiency on ARC-AGI-3 pulls Chollet’s AGI forecast forward

GPT-6 Astra shows mixed benchmark results but beats human efficiency on ARC-AGI-3. Epoch AI scores the model at 169 points, while Artificial Analysis ranks it behind Claude Fable 5.1. The standout performance on ARC-AGI-3 prompts François Chollet to accelerate his AGI timeline due to faster than expected progress.

The Decoder·Sep 4, 2026

Opening Kapyn…