Parallel Codex Accounts Crack 20 Erdős Problems in a Single Run
💡 • Invest in AI orchestration and infrastructure companies that enable multi-account parallelism (e.g., LangChain, Modal). • Launch a side hustle offering 'parallel AI math solving' to quantitative hedge funds, engineering firms, or academic researchers. • Use the technique to scan multiple crypto blockchains simultaneously for arbitrage or vulnerability detection, boosting profit potential. • Apply parallel AI to real estate portfolio risk analysis, evaluating hundreds of properties in minutes instead of days. • Build a custom tool that automates multi-account AI task distribution and sell it as a SaaS product to developers.
A developer used 20 separate OpenAI Codex accounts working in parallel to solve 20 classic Erdős problems simultaneously. The technique demonstrates how distributed AI agents can dramatically accelerate mathematical research, opening new avenues for algorithmic trading, quant finance, and automated theorem proving.
A Hacker News post from Starfleet Math reveals that a single operator deployed 20 Codex accounts in parallel, each tasked with solving one of the 20 Erdős problems—a set of notoriously difficult combinatorial and number theory puzzles. The simultaneous execution suggests that large-scale AI orchestration can now handle multiple complex reasoning tasks at once, bypassing the serial bottlenecks typical of single-instance models.
Parallelizing AI agents for math is not just a novelty; it has direct implications for industries that rely on rapid pattern recognition and optimization. In finance, for example, solving multiple constraint-satisfaction problems concurrently could enable real-time portfolio rebalancing across thousands of scenarios. The same logic applies to logistics, cryptography, and supply chain modeling.
For investors, the breakthrough highlights the growing value of AI orchestration platforms. Companies that build tools to manage multi-account, multi-model workflows—like LangChain, Modal, or Ray—may see increased demand. The ability to farm out tasks to many AI instances cheaply could lower the cost of R&D in quantitative hedge funds and tech firms.
Side hustlers and freelancers can replicate this approach to offer “AI math consultancy” services. By renting compute and subscribing to multiple API accounts, an individual could solve dozens of custom optimization problems for clients in engineering, gaming, or academic publishing. The low barrier to entry (just an API key and a coordination script) makes this a viable niche.
Real estate and crypto markets also stand to benefit. In real estate, parallel AI agents could evaluate hundreds of property portfolios for risk and yield simultaneously. In crypto, they could scan multiple blockchains for arbitrage opportunities or smart contract vulnerabilities in parallel, increasing the speed of profitable trades.
The catch is that existing Codex accounts have rate limits and costs. Running 20 accounts in parallel multiplies the API bill, but the time saved can more than compensate. Early adopters who master this parallelized workflow may gain a temporary edge before the technique becomes commoditized.
Read the full story
Original reporting and related coverage — attribution links only, not paid recommendations.
Trade this story
Partner links — OppHub may earn a commission at no extra cost to you.
Structured tickers, ETFs, hedges, and invalidation triggers from this story — not personalized advice.