Home/Blog/Show HN: Nenya, a 0.5B-active base trained on 60B tokens, rivals Qwen3.5-2BAI & MLShow HN: Nenya, a 0.5B-active base trained on 60B tokens, rivals Qwen3.5-2BHacker News Show2026-11-10 08:010 viewsWould love some feedback! :) Comments URL: https://news.ycombinator.com/item?id=50434208 Points: 1 # Comments: 0Source:https://huggingface.co/ArsalanZ/Nenya-2.5B-A0.5B-BaseMy favoritesPreviousShow HN: Stop Walking – A 10-level, one-button 3D rage game in the browserNextGoogle Antigravity for Knowledge WorkRelated postsWaymo school bus recall: NHTSA opens query into whether the fix workedOriginally published on ColdCaseTech . Disclosure: this post contains affiliate links. We may earn a commission at no ex0dev.to NewsAnthropic Claude Fable 5.2 launches at $8 per million input tokensOriginally published on ColdCaseTech . Disclosure: this post contains affiliate links. We may earn a commission at no ex0dev.to NewsYour coding agent already saved the API key you thought was only in .envCoding agents keep a local history of each session. That history is useful: you can resume yesterday's work, see which t0dev.to ProductivityHow to Choose an AI Model Inference Service: Managed APIs vs ML Endpoints vs Serverless GPUs vs Self-Hosted ServingChoose an inference deployment option by model type (proprietary vs open-weight), traffic shape, latency target, model f0dev.to Machine LearningFrom a 40-Page Research PDF to a 10-Slide Deck, Using Only EmailDisclosure: I'm the developer of Maildoer ( https://maildoer.com ). This post shows how I use it myself. A lot of my wee0dev.to TutorialRemove a Photo Background Without Installing an AppNeed to drop a product photo on a white background, or cut a subject out of a selfie for a profile picture? You don't ne0dev.to Tutorial