Skip to main content
Back to News Hub
🤖OpenAI
October 14, 2026
General AI

How Third Moment Research runs 10,000 agents for 36+ hours

Overview

Third Moment Research is operating 10,000 autonomous agents continuously for 36+ hours. According to OpenAI, the research lab utilizes models such as GPT-6 Astra, GPT-6.1 Sol, Codex, and Daybreak to handle domain challenges like cybersecurity. These systems are designed to retain focus across multiple days while improving through ongoing use.

Key Takeaways

  • Artificial intelligence research firm Third Moment Research has developed infrastructure to execute 10,000 agents simultaneously for 36+ hours.

    To maintain performance across such long operational windows, the organization integrates multiple advanced models, including GPT-6 Astra, GPT-6.1 Sol, Codex, and a cybersecurity tool named Daybreak.

  • For individuals studying agentic systems, this initiative illustrates how combining targeted domain models with durable execution frameworks enables software to refine its performance while running over multiple days.

    Third Moment Research deploys clusters of 10,000 AI agents capable of operating continuously for 36+ hours.

  • The operational framework utilizes specialized models such as GPT-6 Astra, GPT-6.1 Sol, Codex, and Daybreak.

    The lab applies these long-running software agents to complex tasks within cybersecurity.

  • These autonomous systems are engineered to maintain focus over multiple days while improving through practical application.
  • This milestone highlights how modern AI architectures are moving from short prompts toward persistent software agents capable of long-term concentration.

Artificial intelligence research firm Third Moment Research has developed infrastructure to execute 10,000 agents simultaneously for 36+ hours. To maintain performance across such long operational windows, the organization integrates multiple advanced models, including GPT-6 Astra, GPT-6.1 Sol, Codex, and a cybersecurity tool named Daybreak. This milestone highlights how modern AI architectures are moving from short prompts toward persistent software agents capable of long-term concentration.

For individuals studying agentic systems, this initiative illustrates how combining targeted domain models with durable execution frameworks enables software to refine its performance while running over multiple days. Third Moment Research deploys clusters of 10,000 AI agents capable of operating continuously for 36+ hours. The operational framework utilizes specialized models such as GPT-6 Astra, GPT-6.1 Sol, Codex, and Daybreak.

For more details please read the original article at OpenAI.

Continue Learning

Comments

Comments appear only after moderation. Your email identifies your submission to the moderator and is never displayed here.

No approved comments yet.

Originally published by OpenAI
Read the original