OpenAI profiles Third Moment Research, an AI lab running 10,000 agents for more than 36 hours
Announced October 14, 2026
Verified (fact)
OpenAI published a case study on AI lab Third Moment Research. The lab uses GPT-6 Astra, GPT-6.1 Sol, Codex and Daybreak for security to run 10,000 agents for more than 36 hours. The agents are built to focus on a single task for days and improve with use.
Needs fact-check (claim)
OpenAI says the agents stay focused for days and improve as they are used.
Why it matters (interpretation)
Notable as a real-world case of long-running, large-scale parallel agent operation. At this scale, cost and oversight become practical challenges.
Sources
- Official How Third Moment Research runs 10,000 agents for 36+ hours OpenAI
More about OpenAI
-
Eight US states sue OpenAI, saying ChatGPT is not safe for children (report)
Iowa, Kansas, Nebraska, West Virginia, Arkansas, Arizona, Idaho and Montana have each filed lawsuits against OpenAI. They allege the...
-
OpenAI profiles Amp, which uses GPT-6 Astra, Ultrafast, GPT-Live-1 and 'Sign in with ChatGPT' to extend agents beyond coding
OpenAI published a case study on Amp. Amp uses GPT-6 Astra, Ultrafast, GPT-Live-1 and 'Sign in with ChatGPT' to extend agents to work...
-
OpenAI profiles SentinelOne's use of 'Daybreak': malware analysis cut 'from weeks to hours'
According to OpenAI, security company SentinelOne used OpenAI Daybreak to cut malware analysis that took weeks down to hours, freeing...
-
OpenAI profiles Block's AI security build-out: faster threat investigation and patch suggestions in minutes
OpenAI described how Block is building an AI-based security system with its technology. Block's security team investigates threats...
-
OpenAI publishes analysis of agentic AI adoption among EU businesses and small firms
OpenAI released an analysis of how agentic AI adoption is growing among EU businesses and small firms, and where deeper use could create...