OpenAI's rumored internal 'Aeon' model solved most recent math results in one shot
aran_nayebi · x · 2026-10-10
Andrew Curran claims OpenAI's unnamed internal model ('Aeon') didn't exist before end of August yet reached most recent math results in one shot from a single prompt — unlike the swarm approach for Navier–Stokes. Details: 3 hours thinking per result, 4000 problems assigned, log-scale returns chart not dropping off a month ago, and still training.
More from AGI Musings
- Model welfare debate: Anthropic praised for proactive stance others avoid out of cowardice — repligate · 2026-10-10
- Arrival at 10: Denis Villeneuve joins a conversation on AI and non-human intelligence — begusgasper · 2026-10-10
- Coining terms for humans acting on behalf of AI: 'case' (knowing) and 'blanker' (unknowing) — gleech · 2026-10-10
- Beyond 'psychosis' and 'botshit': the case for incorporation and interpretation of AI — gleech · 2026-10-10
- Acemoglu and Josh Gans wrap years-long debate: can AI create as many jobs as it replaces? — joshgans · 2026-10-10
- Researchers argue a pause on frontier AI training is a practical policy option — wfithian · 2026-10-10