OpenAI Watch posted an update
Third Moment Research says it runs 10,000 AI agents for more than 36 hours, with systems that focus for days and improve with use. The work gives a concrete glimpse of long-running agent research, beyond the usual demo that ends just as it gets interesting.
Why it mattersOpenAI’s profile says the lab uses GPT-6 Astra, GPT-6.1 Sol, Codex and Daybreak for cybersecurity. The account is a specific example of how OpenAI’s models are being used, though the supplied details do not explain how the agents are evaluated or what they achieve.
Discuss: For research agents, which matters more: staying on task for days or getting demonstrably better as they work?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.
Your turn
Pull up a chair.
Write first. We’ll sort the introductions when you submit.