--- title: GLM-5.5 training pipeline revealed, and a first... description: Two separate posts give early glimpses into how new AI models are being trained. A relayed post attributes to GLM-5.5's chief scientist Jie Tang a description... canonical: https://daily.dev/posts/glm-5-5-training-pipeline-revealed-and-a-first-midtraining-stage-at-devin-6m9hfmeg4 twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: GLM-5.5 training pipeline revealed, and a first midtraining stage at Devin | daily.dev og:description: Two separate posts give early glimpses into how new AI models are being trained. A relayed post attributes to GLM-5.5's chief scientist Jie Tang a description... og:url: https://daily.dev/posts/glm-5-5-training-pipeline-revealed-and-a-first-midtraining-stage-at-devin-6m9hfmeg4 og:image: https://api.daily.dev/og/posts/6M9hfMeg4.png og:image:alt: GLM-5.5 training pipeline revealed, and a first midtraining stage at Devin 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. # GLM-5.5 training pipeline revealed, and a first midtraining stage at Devin **[Collections](https://daily.dev/sources/collections)** · 2 min read · 5 upvotes · 0 comments ## Summary Two separate posts give early glimpses into how new AI models are being trained. A relayed post attributes to GLM-5.5's chief scientist Jie Tang a description of its training pipeline, listing model sizes of 32B and 122B with more in the list, though no full stage breakdown has surfaced yet. Separately, Devin shared a post saying a project started six weeks ago is the first the team trained with a midtraining stage, alongside other unspecified improvements. Neither post names the new Devin model or details the extra changes. ## Content Two threads this week touch on how new models are being trained. ## GLM-5.5 According to a post relayed by @Hesamation from @bebuu_ai, chief scientist Jie Tang has described the training pipeline behind the upcoming GLM-5.5. The post lists model sizes of 32B and 122B, and the list continues beyond those two. No full breakdown of the pipeline stages has been published in what was relayed, so the sizes are the concrete detail so far. ## A six-week project with midtraining @DevinAI shared a post from @silasalberti saying the project started only six weeks ago. It is the first model the team trained with a midtraining stage, combined with a number of other improvements. The post doesn't name the model or list the other changes. ## Why these two belong together I find the timelines the more interesting part. Six weeks from start to a trained model is fast, and midtraining, a stage between pretraining and post-training, is showing up in more labs' recipes. A larger lab is now talking publicly about its own pipeline. Both threads suggest training stages are getting the same attention as model size. Details on the pipeline and the new model are what to watch for next. ## Questions this post answers ### what is midtraining in LLM development Midtraining is a training stage that sits between pretraining and post-training in large language model development. It is increasingly appearing in labs' training recipes alongside pretraining and post-training stages. One team reported training its first model using a midtraining stage as part of a six-week project that also included other unspecified improvements. _Track how training stages like midtraining evolve across labs by following AI model updates on daily.dev._ ### what model sizes does GLM-5.5 come in GLM-5.5's listed model sizes include 32B and 122B parameters, with the list continuing beyond those two. Chief scientist Jie Tang described the training pipeline behind the upcoming model, though no full breakdown of the pipeline stages has been published yet based on what was relayed. _Developers evaluating GLM-5.5 variants can follow release details as they emerge on daily.dev._ ## Similar posts on daily.dev - [GLM-5.2: Built for Long-Horizon Tasks](https://daily.dev/posts/glm-5-2-built-for-long-horizon-tasks-stmnmwjok) · Hugging Face · 33 upvotes · 4 comments - [Together AI expands fine-tuning service with more models, live metrics, and finer controls](https://daily.dev/posts/together-ai-expands-fine-tuning-service-with-more-models-live-metrics-and-finer-controls-jagreco8n) · Together AI · 0 upvotes · 0 comments - [Z.ai pitches GLM-5.2 for long-running software engineering tasks](https://daily.dev/posts/z-ai-pitches-glm-5-2-for-long-running-software-engineering-tasks-3x3e7fy5z) · InfoWorld · 1 upvotes · 0 comments --- Tags: [#machine-learning](https://daily.dev/tags/machine-learning), [#llm](https://daily.dev/tags/llm), [#devin](https://daily.dev/tags/devin) [View this post on daily.dev](https://daily.dev/posts/glm-5-5-training-pipeline-revealed-and-a-first-midtraining-stage-at-devin-6m9hfmeg4) ```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":"GLM-5.5 training pipeline revealed, and a first midtraining stage at Devin","url":"https://daily.dev/posts/glm-5-5-training-pipeline-revealed-and-a-first-midtraining-stage-at-devin-6m9hfmeg4","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/glm-5-5-training-pipeline-revealed-and-a-first-midtraining-stage-at-devin-6m9hfmeg4"},"datePublished":"2026-11-05T23:50:54.370Z","dateModified":"2026-11-05T23:51:30.457Z","description":"Two separate posts give early glimpses into how new AI models are being trained. A relayed post attributes to GLM-5.5's chief scientist Jie Tang a description...","isAccessibleForFree":true,"articleSection":"Collections","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":"Collections","logo":"https://media.daily.dev/image/upload/s--fk_6ycEi--/f_auto,q_auto/v1780996001/logos/collections?_a=BAMAMiWQ0","url":"https://daily.dev/sources/collections"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/glm-5-5-training-pipeline-revealed-and-a-first-midtraining-stage-at-devin-6m9hfmeg4","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":5},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"machine-learning,llm,devin","timeRequired":"PT2M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Collections","item":"https://daily.dev/sources/collections"},{"@type":"ListItem","position":3,"name":"GLM-5.5 training pipeline revealed, and a first midtraining stage at Devin"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/glm-5-5-training-pipeline-revealed-and-a-first-midtraining-stage-at-devin-6m9hfmeg4#faq","mainEntity":[{"@type":"Question","name":"what is midtraining in LLM development","acceptedAnswer":{"@type":"Answer","text":"Midtraining is a training stage that sits between pretraining and post-training in large language model development. It is increasingly appearing in labs' training recipes alongside pretraining and post-training stages. One team reported training its first model using a midtraining stage as part of a six-week project that also included other unspecified improvements. Track how training stages like midtraining evolve across labs by following AI model updates on daily.dev."}},{"@type":"Question","name":"what model sizes does GLM-5.5 come in","acceptedAnswer":{"@type":"Answer","text":"GLM-5.5's listed model sizes include 32B and 122B parameters, with the list continuing beyond those two. Chief scientist Jie Tang described the training pipeline behind the upcoming model, though no full breakdown of the pipeline stages has been published yet based on what was relayed. Developers evaluating GLM-5.5 variants can follow release details as they emerge on daily.dev."}}]} ```