--- title: RLMs are the new buzzword, and the people who get it are... description: RLM (recursive language model) is an emerging term in AI developer circles referring to a supervised fine-tuned model running inside a REPL environment. Unlike... canonical: https://daily.dev/posts/rlms-are-the-new-buzzword-and-the-people-who-get-it-are-being-annoyingly-coy-about-why-40lzufxx9 twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: RLMs are the new buzzword, and the people who get it are being annoyingly coy about why | daily.dev og:description: RLM (recursive language model) is an emerging term in AI developer circles referring to a supervised fine-tuned model running inside a REPL environment. Unlike... og:url: https://daily.dev/posts/rlms-are-the-new-buzzword-and-the-people-who-get-it-are-being-annoyingly-coy-about-why-40lzufxx9 og:image: https://api.daily.dev/og/posts/40LzuFXx9.png og:image:alt: RLMs are the new buzzword, and the people who get it are being annoyingly coy about 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. # RLMs are the new buzzword, and the people who get it are being annoyingly coy about why **[Trends](https://daily.dev/sources/trends)** · 2 min read · 1 upvotes · 0 comments ## Summary RLM (recursive language model) is an emerging term in AI developer circles referring to a supervised fine-tuned model running inside a REPL environment. Unlike standard LLM agents that simulate recursion by unrolling it into sequential steps, an RLM can write and execute genuinely recursive programs within a real execution environment. DeepAgents added RLM support recently, and the concept is gaining traction with figures like Harrison Chase (LangChain) amplifying discussion. The discourse is still early-stage, with knowledgeable insiders being deliberately vague about the implications — a pattern that could signal a genuine capability unlock for code generation and reasoning tasks, or simply pre-hype cycle behavior. ## Content "Recursive language model" is the term floating around developer Twitter this week, and if you missed it, here's the short version: an RLM is a setup where a language model can call itself recursively, typically inside a REPL environment where it can write and execute real recursive programs that actually recurse at runtime. The honest reaction from people who've been in this space a while: this isn't new. Sydney Runkle pointed out that @a1zhang wrote the original RLM harness before the term started trending. The deepagents team added RLM support months ago. The concept has been sitting there; the branding just caught up. That said, the "not new" crowd isn't fully winning the argument. willccbb pushed back on the dismissiveness: RLMs at depth=1 can do *true* recursion because you're in a REPL, meaning the model isn't simulating recursion in its output, it's writing programs that actually recurse when run. That's a real distinction worth caring about. Elnashar (omarsar0) is being deliberately cryptic: "there is something particularly interesting about RLMs and people are about to find out why." Which is either a genuine signal that something is coming, or the classic move of sounding informed without saying anything. The meta here is familiar. A concept gets a catchy name, the name spreads, people who built the thing six months ago get annoyed, and then someone points out the name actually does capture something real. The discourse is currently somewhere between steps two and three. What's worth watching: if RLMs are genuinely useful because they enable real recursive execution rather than just recursive-looking outputs, that's a meaningful capability gap from standard agentic setups. The harness framing suggests this is becoming infrastructure, not just a demo trick. ## Questions this post answers ### What is an RLM (recursive language model) and how does it differ from a regular LLM agent? An RLM (recursive language model) is a supervised fine-tuned model running inside a REPL environment, allowing it to write and execute genuinely recursive programs rather than simulating recursion in a single forward pass. Standard LLM agents fake recursion by unrolling it into sequential steps, whereas an RLM in a REPL can write a function that calls itself and actually run it — a meaningful difference for code generation and reasoning tasks. _Engineers building or evaluating agent frameworks track emerging capability distinctions like this on daily.dev._ ## Similar posts on daily.dev - [Recursive Language Models: An All-in-One Deep Dive](https://daily.dev/posts/recursive-language-models-an-all-in-one-deep-dive-bx3jpqplg) · Towards Data Science · 2 upvotes · 0 comments - [Everything You Need to Know About Recursive Language Models](https://daily.dev/posts/everything-you-need-to-know-about-recursive-language-models-ujzedkour) · Machine Learning Mastery · 2 upvotes · 0 comments - [Recursive Language Models](https://daily.dev/posts/recursive-language-models-jnp1ro58a) · Hacker News · 3 upvotes · 0 comments - [\[2512.24601\] Recursive Language Models](https://daily.dev/posts/2512-24601-recursive-language-models-k5s8xjxqv) · Hacker News · 6 upvotes · 0 comments --- Tags: [#llm](https://daily.dev/tags/llm), [#ai-agents](https://daily.dev/tags/ai-agents), [#langchain](https://daily.dev/tags/langchain) [View this post on daily.dev](https://daily.dev/posts/rlms-are-the-new-buzzword-and-the-people-who-get-it-are-being-annoyingly-coy-about-why-40lzufxx9) ```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":"RLMs are the new buzzword, and the people who get it are being annoyingly coy about why","url":"https://daily.dev/posts/rlms-are-the-new-buzzword-and-the-people-who-get-it-are-being-annoyingly-coy-about-why-40lzufxx9","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/rlms-are-the-new-buzzword-and-the-people-who-get-it-are-being-annoyingly-coy-about-why-40lzufxx9"},"datePublished":"2026-08-07T21:35:47.941Z","dateModified":"2026-08-09T13:37:52.106Z","description":"RLM (recursive language model) is an emerging term in AI developer circles referring to a supervised fine-tuned model running inside a REPL environment. Unlike...","isAccessibleForFree":true,"articleSection":"Trends","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":"Trends","logo":"https://media.daily.dev/image/upload/s--ZfSp3asX--/f_auto,q_auto/v1780996004/logos/trends?_a=BAMAMiWQ0","url":"https://daily.dev/sources/trends"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/rlms-are-the-new-buzzword-and-the-people-who-get-it-are-being-annoyingly-coy-about-why-40lzufxx9","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"llm,ai-agents,langchain","timeRequired":"PT2M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Trends","item":"https://daily.dev/sources/trends"},{"@type":"ListItem","position":3,"name":"RLMs are the new buzzword, and the people who get it are being annoyingly coy about why"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/rlms-are-the-new-buzzword-and-the-people-who-get-it-are-being-annoyingly-coy-about-why-40lzufxx9#faq","mainEntity":[{"@type":"Question","name":"What is an RLM (recursive language model) and how does it differ from a regular LLM agent?","acceptedAnswer":{"@type":"Answer","text":"An RLM (recursive language model) is a supervised fine-tuned model running inside a REPL environment, allowing it to write and execute genuinely recursive programs rather than simulating recursion in a single forward pass. Standard LLM agents fake recursion by unrolling it into sequential steps, whereas an RLM in a REPL can write a function that calls itself and actually run it — a meaningful difference for code generation and reasoning tasks. Engineers building or evaluating agent frameworks track emerging capability distinctions like this on daily.dev."}}]} ```