SIGNAL / RESEARCH

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

SOURCEOpenAI
PUBLISHEDOctober 14, 2026
LAYERResearch
SIGNAL SCORE53/100

SOURCE SUMMARY

With GPT-6 Astra and GPT-6.1 Sol, Codex, and Daybreak for cybersecurity, this AI lab builds agents that focus for days and improve with use.

WHY IT MATTERS

This is primarily a deployment or customer-use signal. It is useful for understanding where the technology is being applied, but the result should not be generalized beyond the described workflow without comparable evidence.

WHAT TO VERIFY

Check the customer’s baseline, workflow scope, measurement period, model configuration, human involvement, cost accounting, and whether the reported outcome was independently evaluated.

TOPICSOpenAIAI Agents
MODELSNo named model detected