Third Moment Research runs 10,000 Agents for 36+ hours with GPT-6 Astra and GPT-6.1 Sol

Third Moment Research demonstrated its Agent runtime: on top of GPT-6 Astra and GPT-6.1 Sol, combined with Codex and the cybersecurity-focused Daybreak, it kept roughly 10,000 Agents running continuously for over 36 hours. The description emphasizes that these Agents stay focused for long stretches and improve as they are used. This scale is directly relevant to teams building multi-Agent systems: per-Agent context retention and failure recovery, not concurrency itself, are the main bottlenecks for sustaining 36-hour continuous operation. For teams handling security automation, Daybreak being part of the same runtime means persistent security Agent watch duty is an explicitly covered scenario.