Third Moment Research used GPT‑6 models to keep ten thousand agents active for more than a day and a half.
NoteBy OossaPublished by Oossa: 1 min read
Third Moment Research, an OpenAI lab, kept 10,000 AI agents running continuously for more than 36 hours. The agents used the new GPT‑6 Astra and GPT‑6.1 Sol models, plus Codex and Daybreak for cybersecurity tasks. The agents stayed focused for days and improved with use, according to the lab’s description.
It shows that large‑scale, long‑running AI agents are now feasible, opening the door to persistent automation in business and security.
Tools & agents
Google DeepMind engineers report that AI writes most production code but still needs human review and shows no sign of accelerating AI research.
Tools & agents
Shopify now allows shoppers to link any AI assistant to their account so the bot can browse and purchase from over a billion listings.
Tools & agents
Facebook now offers a feature called Your Algorithm that lets users mark topics as “interested” or “not interested” and see changes in their feed right away.
Oossa · Newsletter
The week in AI, explained
Every Monday: the stories worth knowing, in plain language. Free, no spam.
