[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-effgen-lets-small-language-models-run-as-local-agents":10},{"siteName":4,"siteTagline":5,"publisherName":4,"contactEmail":6},"The Revision","Tech news, decoded.","editor@therevision.news",{"gaMeasurementId":8,"adsenseClientId":9},"G-ZW2MV82GYR","ca-pub-8533917693782264",{"article":11},{"id":12,"slug":13,"title":14,"dek":15,"body_md":16,"tags_json":17,"published_at":18,"created_at":19,"updated_at":20,"status":21,"review_note":22,"review_notes":23,"image_url":22,"persona_id":22,"persona_name":22,"section":22,"tags":24,"sources":28,"feedback":32,"feedback_at":22,"cost_usd":32,"total_tokens":32},1349,"effgen-lets-small-language-models-run-as-local-agents","EffGen lets small language models run as local agents","Open-source EffGen trims prompts, splits tasks and routes work, letting compact models act autonomously without cloud APIs.","EffGen lets tiny language models act as on‑device agents.\n\nThe paper introduces EffGen, a framework built for small language models (SLMs). It shrinks tool‑calling prompts by 70‑80% (57% on average), breaks complex queries into dependent subtasks, routes work using five complexity factors, and offers a unified memory that mixes short‑term, long‑term and vector stores. The system also speaks multiple agent protocols in a single API. Benchmarks across 13 tasks show higher success rates and faster execution than LangChain, AutoGen and Smolagents, while using less memory.\n\nThis matters because most current agentic pipelines rely on large, cloud‑hosted models whose API fees and data‑privacy risks are growing concerns. By proving that prompt compression helps SLMs more and that routing helps larger models, EffGen gives developers a cost‑effective path to deploy autonomous AI locally, expanding use cases in regulated industries.\n\nThe code is Apache‑2.0, available on GitHub, and joins a modest but rising pool of open‑source tools that aim to democratize agentic AI.","[\"small-language-models\",\"agent-framework\",\"open-source\"]","2026-06-16T04:00:00.000Z","2026-06-17T05:17:03.828Z","2026-06-17T05:17:06.737Z","published",null,[],[25,26,27],"small-language-models","agent-framework","open-source",[29],{"name":30,"url":31},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2602.00887",0]