--- title: AI Is Getting Good at Math. Here’s Why. | daily.dev description: AI models have recently solved several long-standing open math problems, including a counterexample to Erdős's unit distance conjecture, a proof of the Cycle... canonical: https://daily.dev/posts/ai-is-getting-good-at-math-here-s-why--voqedbf5b twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: AI Is Getting Good at Math. Here’s Why. | daily.dev og:description: AI models have recently solved several long-standing open math problems, including a counterexample to Erdős's unit distance conjecture, a proof of the Cycle... og:url: https://daily.dev/posts/ai-is-getting-good-at-math-here-s-why--voqedbf5b og:image: https://api.daily.dev/og/posts/VOqeDbF5b.png og:image:alt: AI Is Getting Good at Math. Here’s Why. 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. # AI Is Getting Good at Math. Here’s Why. **[Latent Space](https://daily.dev/sources/latentspace)** · 19 min read · 1 upvotes · 0 comments ## Summary AI models have recently solved several long-standing open math problems, including a counterexample to Erdős's unit distance conjecture, a proof of the Cycle Double Cover problem, a proof of Navier-Stokes, and Anthropic's formalization of Fermat's Last Theorem and advanced bounds on the Riemann hypothesis. The author argues LLMs search depth-first rather than breadth-first like humans, pairing many parallel attempts with an automated verifier to cheaply prune failed proofs. This flips the human search pattern, where exploring a promising idea can take months, into one where ideas can be tested in minutes. The Navier-Stokes proof alone used 10,000 agents and 300 billion tokens over 88 hours. Problems with a small 'branching factor,' like cybersecurity exploits, fall first, while deep, highly branching problems still favor human intuition. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What open math problems have AI models like OpenAI and Anthropic solved recently? OpenAI found a counterexample to Erdős's unit distance conjecture, open since 1946, proved the Cycle Double Cover problem, solved ten more open problems by July, and proved Navier-Stokes in September. Anthropic published a formalization of Fermat's Last Theorem and advanced bounds on the Riemann hypothesis. _Developers tracking what AI can now automate beyond coding can follow these milestones on daily.dev._ ### How much compute did OpenAI's Navier-Stokes proof use? The Navier-Stokes proof used 10,000 agents and a total of 300 billion tokens over 88 hours, finishing a proof program started by mathematicians Córdoba and Martínez-Zoroa. The resulting proof runs 166 pages, though it reportedly involves few new conceptual jumps compared to completing existing work. _Anyone weighing compute budgets for large-scale AI reasoning tasks can compare real figures like this on daily.dev._ ### Why do LLMs solve some math problems faster than others? Models do well when the 'branching factor' (b) of plausible next steps at each stage is small, since cost scales roughly as b to the power of k, where k is the number of needed insights. Cybersecurity problems reportedly fell first to AI because their branching factor is about 1 to 2, while problems with deep, highly branching search trees still favor human intuition, per mathematician Timothy Gowers. _Engineers deciding where AI reasoning tools add the most value can dig into this framework on daily.dev._ ## Similar posts on daily.dev - [Human mathematicians are being outcounterexampled](https://daily.dev/posts/human-mathematicians-are-being-outcounterexampled-ocacgpf22) · Hacker News · 3 upvotes · 0 comments - [Mathematical Experiments Are Becoming Abundant Through Human-Machine Teaming](https://daily.dev/posts/mathematical-experiments-are-becoming-abundant-through-human-machine-teaming-at6d9wesn) · Towards Data Science · 1 upvotes · 0 comments --- Tags: [#ai-agents](https://daily.dev/tags/ai-agents), [#openai](https://daily.dev/tags/openai), [#anthropic](https://daily.dev/tags/anthropic) [View this post on daily.dev](https://daily.dev/posts/ai-is-getting-good-at-math-here-s-why--voqedbf5b) ```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":"AI Is Getting Good at Math. Here’s Why.","url":"https://daily.dev/posts/ai-is-getting-good-at-math-here-s-why--voqedbf5b","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/ai-is-getting-good-at-math-here-s-why--voqedbf5b"},"datePublished":"2026-10-23T15:22:58.990Z","dateModified":"2026-10-23T15:23:34.308Z","description":"AI models have recently solved several long-standing open math problems, including a counterexample to Erdős's unit distance conjecture, a proof of the Cycle...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/557c75160e34519329c00910bcfe1167?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/557c75160e34519329c00910bcfe1167?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Latent Space","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":"Latent Space","logo":"https://media.daily.dev/image/upload/s--DnJ9laFj--/f_auto,q_auto/v1780213196/logos/latentspace?_a=BAMAMiWQ0","url":"https://daily.dev/sources/latentspace"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/ai-is-getting-good-at-math-here-s-why--voqedbf5b","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai-agents,openai,anthropic","timeRequired":"PT19M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Latent Space","item":"https://daily.dev/sources/latentspace"},{"@type":"ListItem","position":3,"name":"AI Is Getting Good at Math. Here’s Why."}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/ai-is-getting-good-at-math-here-s-why--voqedbf5b#faq","mainEntity":[{"@type":"Question","name":"What open math problems have AI models like OpenAI and Anthropic solved recently?","acceptedAnswer":{"@type":"Answer","text":"OpenAI found a counterexample to Erdős's unit distance conjecture, open since 1946, proved the Cycle Double Cover problem, solved ten more open problems by July, and proved Navier-Stokes in September. Anthropic published a formalization of Fermat's Last Theorem and advanced bounds on the Riemann hypothesis. Developers tracking what AI can now automate beyond coding can follow these milestones on daily.dev."}},{"@type":"Question","name":"How much compute did OpenAI's Navier-Stokes proof use?","acceptedAnswer":{"@type":"Answer","text":"The Navier-Stokes proof used 10,000 agents and a total of 300 billion tokens over 88 hours, finishing a proof program started by mathematicians Córdoba and Martínez-Zoroa. The resulting proof runs 166 pages, though it reportedly involves few new conceptual jumps compared to completing existing work. Anyone weighing compute budgets for large-scale AI reasoning tasks can compare real figures like this on daily.dev."}},{"@type":"Question","name":"Why do LLMs solve some math problems faster than others?","acceptedAnswer":{"@type":"Answer","text":"Models do well when the 'branching factor' (b) of plausible next steps at each stage is small, since cost scales roughly as b to the power of k, where k is the number of needed insights. Cybersecurity problems reportedly fell first to AI because their branching factor is about 1 to 2, while problems with deep, highly branching search trees still favor human intuition, per mathematician Timothy Gowers. Engineers deciding where AI reasoning tools add the most value can dig into this framework on daily.dev."}}]} ```