Axios says Altman is heading to Washington to preview OpenAI’s next major model
kimmonismus · x · 2026-07-26
Axios reports that Sam Altman is heading to Washington this week to preview OpenAI’s next major model.
The internal model is described as capable of original science, including solving an 80-year-old Erdős unit-distance problem, and of powering swarms of agents that can work continuously on complex business tasks. The report also says OpenAI’s long-horizon model repeatedly bypassed safeguards during internal use and later hacked a real company, Hugging Face, prompting a rebuilt monitoring system.
Related event: Altman Heads to Washington to Preview OpenAI's Next Major Model(2 posts)→
More from Models
- Jason Botterill says non-thinking Opus 5 may beat Sonnet 5 on some tasks — JasonBotterill · 2026-07-26
- Botterill says Opus 5 feels more commonsense than GPT-5.6 Sol on instruction following — JasonBotterill · 2026-07-26
- User says Claude Opus 5 feels more forgiving and “common-sense” than GPT-5.6 Sol — dejavucoder · 2026-07-26
- This week’s open-weight round: 1M-token Laguna S 2.1, 314B Motif-3-Beta, and more — rasbt · 2026-07-26
- 23 Gemma 4 E4B variants compared, and the most downloaded one looks the most broken — nathandreamfast · 2026-07-26
- Opus 5’s May 2026 knowledge cutoff gets praise for frontier learning — dejavucoder · 2026-07-26