High impact

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

What happened

Third Moment Research runs 10,000 AI agents simultaneously for 36‑48 hour investigations using OpenAI’s GPT‑6 Astra, GPT‑6 Sol, GPT‑6.1 Sol and the Codex harness. The system creates synthetic scenarios to spot weaknesses, then improves agents automatically. In quantitative finance, Codex identified a backtesting bottleneck and, after optimization, cut runtime from about three hours to three‑to‑five minutes, saving tokens and context. Agents operate in Dockerized environments with terabytes of data, using subagents and outer harnesses to retry or escalate unfinished tasks. The approach lets agents design and run hundreds of parallel experiments to refine trading strategies.

Why it matters

For readers, this shows how AI can dramatically speed up complex financial research, reducing costs and improving productivity. It also illustrates a model for self‑improving AI systems that could be applied to other specialized tasks.

Reality check

An automated, sceptical read of the source

Mostly marketingIncrementalNotableGenuine breakthrough

Incremental. The article describes scaling and efficiency gains within an existing AI framework rather than a fundamentally new technology.

Who it affectsDevelopersBusinessesResearchers

What to watchWatch for further extensions of the self‑improving agent framework into other domains beyond finance.

Do you agree with this rating?Agree Disagree

More from OpenAI

Introducing ChatGPT WalletOpenAI · 13 Nov 2026Shield helps technicians hand off legacy portal work with OpenAIOpenAI · 13 Nov 2026OpenAI launches ChatGPT Wallet, AI agents pay for purchaseOpenAI · 13 Nov 2026OpenAI disbands its robotics team as its general models start running robotsOpenAI · 13 Nov 2026Yang-Mills, OpenAI proof, and the mass gap problemOpenAI · 13 Nov 2026

This page was collected from openai.com and summarised automatically by AIPulsee's pipeline. See how it runs.