--- title: How Third Moment Research runs 10,000 agents for 36+ hours description: Third Moment Research, a quantitative finance startup, runs up to 10,000 autonomous AI agents concurrently on research sessions lasting 36-48 hours using... canonical: https://daily.dev/posts/how-third-moment-research-runs-10-000-agents-for-36-hours-ummtnlk1q twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: How Third Moment Research runs 10,000 agents for 36+ hours | daily.dev og:description: Third Moment Research, a quantitative finance startup, runs up to 10,000 autonomous AI agents concurrently on research sessions lasting 36-48 hours using... og:url: https://daily.dev/posts/how-third-moment-research-runs-10-000-agents-for-36-hours-ummtnlk1q og:image: https://api.daily.dev/og/posts/UmMtNlK1Q.png og:image:alt: How Third Moment Research runs 10,000 agents for 36+ hours og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # How Third Moment Research runs 10,000 agents for 36+ hours **[OpenAI](https://daily.dev/sources/openai)** · 6 min read · 0 upvotes · 0 comments ## Summary Third Moment Research, a quantitative finance startup, runs up to 10,000 autonomous AI agents concurrently on research sessions lasting 36-48 hours using OpenAI's GPT-6 Astra, GPT-6 Sol, GPT-6.1 Sol models alongside Codex. The agents investigate trading strategies, with Codex itself identifying a caching and memory-packing optimization that cut backtest time from roughly 3 hours to 3-5 minutes (a 97% reduction). The team also uses OpenAI's Daybreak cybersecurity platform to automate penetration testing of the sandboxed research environments before launching simulations at scale. The company plans to scale toward hundreds of thousands of collaborating subagents and trillions of tokens of research. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### How much did Codex reduce backtesting time for a quantitative trading strategy? Codex cut backtesting time for a trading strategy across a year of order-level US equity data from roughly 3 hours down to approximately 3 to 5 minutes, a 97% reduction. It achieved this by identifying a caching optimization and a memory-packing technique after profiling the backtest and analyzing its inputs and usage, which also freed up token budget previously wasted on status polling. _daily.dev surfaces real-world benchmarks like this for teams evaluating agentic coding tools for performance work._ ### How many AI agents can run concurrently in an autonomous research session with GPT-6 Astra and GPT-6.1 Sol? Up to 10,000 agents can run concurrently on investigations lasting 36 to 48 hours, supported by Codex, GPT-6 Astra, GPT-6 Sol, and GPT-6.1 Sol. GPT-6.1 Sol is described as maintaining a coherent thread of computational research across days of compute without losing track of the task, and is noted as the most token-efficient of the models in use. _developers scoping large agent fleets can follow coverage of these deployment patterns on daily.dev._ ### How is agent security tested when running thousands of autonomous research agents at scale? The Daybreak cybersecurity platform is used to automate penetration testing of research environments before simulations launch, checking whether an agent could break out of its sandbox and access the wider network. This automated testing is described as essential for validating infrastructure security at a scale of 10,000 concurrent agents, something not feasible to test manually. _teams securing autonomous agent infrastructure can track these practices through daily.dev._ ## Similar posts on daily.dev - [How V7 gives AI agents institutional memory](https://daily.dev/posts/how-v7-gives-ai-agents-institutional-memory-qelllmnld) · OpenAI · 8 upvotes · 0 comments - [OpenAI gives 100,000 scientists free AI, and shows how](https://daily.dev/posts/openai-gives-100-000-scientists-free-ai-and-shows-how-pyma6bja0) · The Next Web · 0 upvotes · 0 comments - [GPT-6 Astra: The AI That Doesn’t Just Answer. It Gets the Job Done](https://daily.dev/posts/gpt-6-astra-the-ai-that-doesn-t-just-answer-it-gets-the-job-done-zotuejbng) · Medium · 0 upvotes · 0 comments --- Tags: [#ai-agents](https://daily.dev/tags/ai-agents), [#openai](https://daily.dev/tags/openai), [#ai-coding](https://daily.dev/tags/ai-coding), [#openai-codex](https://daily.dev/tags/openai-codex) [View this post on daily.dev](https://daily.dev/posts/how-third-moment-research-runs-10-000-agents-for-36-hours-ummtnlk1q) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"How Third Moment Research runs 10,000 agents for 36+ hours","url":"https://daily.dev/posts/how-third-moment-research-runs-10-000-agents-for-36-hours-ummtnlk1q","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/how-third-moment-research-runs-10-000-agents-for-36-hours-ummtnlk1q"},"datePublished":"2026-10-14T15:30:03.169Z","dateModified":"2026-10-14T15:30:29.318Z","description":"Third Moment Research, a quantitative finance startup, runs up to 10,000 autonomous AI agents concurrently on research sessions lasting 36-48 hours using...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/0adc838ef85080fddb3d950c10dc828d?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/0adc838ef85080fddb3d950c10dc828d?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"OpenAI","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"OpenAI","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/984b3632316f4afaa3d5213d5c64fe65","url":"https://daily.dev/sources/openai"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/how-third-moment-research-runs-10-000-agents-for-36-hours-ummtnlk1q","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai-agents,openai,ai-coding,openai-codex","timeRequired":"PT6M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"OpenAI","item":"https://daily.dev/sources/openai"},{"@type":"ListItem","position":3,"name":"How Third Moment Research runs 10,000 agents for 36+ hours"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/how-third-moment-research-runs-10-000-agents-for-36-hours-ummtnlk1q#faq","mainEntity":[{"@type":"Question","name":"How much did Codex reduce backtesting time for a quantitative trading strategy?","acceptedAnswer":{"@type":"Answer","text":"Codex cut backtesting time for a trading strategy across a year of order-level US equity data from roughly 3 hours down to approximately 3 to 5 minutes, a 97% reduction. It achieved this by identifying a caching optimization and a memory-packing technique after profiling the backtest and analyzing its inputs and usage, which also freed up token budget previously wasted on status polling. daily.dev surfaces real-world benchmarks like this for teams evaluating agentic coding tools for performance work."}},{"@type":"Question","name":"How many AI agents can run concurrently in an autonomous research session with GPT-6 Astra and GPT-6.1 Sol?","acceptedAnswer":{"@type":"Answer","text":"Up to 10,000 agents can run concurrently on investigations lasting 36 to 48 hours, supported by Codex, GPT-6 Astra, GPT-6 Sol, and GPT-6.1 Sol. GPT-6.1 Sol is described as maintaining a coherent thread of computational research across days of compute without losing track of the task, and is noted as the most token-efficient of the models in use. developers scoping large agent fleets can follow coverage of these deployment patterns on daily.dev."}},{"@type":"Question","name":"How is agent security tested when running thousands of autonomous research agents at scale?","acceptedAnswer":{"@type":"Answer","text":"The Daybreak cybersecurity platform is used to automate penetration testing of research environments before simulations launch, checking whether an agent could break out of its sandbox and access the wider network. This automated testing is described as essential for validating infrastructure security at a scale of 10,000 concurrent agents, something not feasible to test manually. teams securing autonomous agent infrastructure can track these practices through daily.dev."}}]} ```