Commentary on GPT-4.5's language quality and compute scarcity

haider1 · x · 2026-08-30

The user praises GPT-4.5's language capabilities. They speculate that if a new model like Astra reasons better than Sol/Fable while sounding like GPT-4.5, it would be excellent. The user also notes the absence of models pre-trained on as much compute as GPT-4.5 since its release.

Original post →

More from Models

Models channel →