AnalysisAI ModelsNovember 14, 2026

Ex-Anthropic researcher Jacob Coxon cites superexponential training gains

Read original source →reddit.com

Jacob Coxon, formerly of OpenAI and Anthropic, says Anthropic's IPO graph shows superexponential algorithmic progress and that Mythos was a point on that curve, not a one-time deviation. He claims Mythos could already be retrained with a fraction of the compute originally needed.

People · Jacob Coxon

2 sources

More stories today

Open the live feed