[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-cutting-ui-code-gen-costs-in-half-with-token-compression":10,"sections":34},{"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":24,"tags":25,"sources":29,"feedback":33,"feedback_at":22,"cost_usd":33,"total_tokens":33},2877,"cutting-ui-code-gen-costs-in-half-with-token-compression","Cutting UI Code Gen Costs in Half With Token Compression","A new open-source framework trims nearly half the compute from AI-driven screenshot-to-code workflows without degrading output quality.","Researchers have published a compression framework that makes AI-generated UI code dramatically cheaper to produce.\n\nEfficientUICoder targets a specific inefficiency in multimodal AI workflows: turning a screenshot of a webpage into working HTML and CSS requires feeding huge numbers of image tokens in and generating huge amounts of code tokens out. The paper's authors found significant redundancy on both ends. Their framework attacks it three ways — detecting UI element regions to preserve layout structure, pruning low-attention image tokens while rescuing high-attention ones from overlooked areas, and penalizing repetitive HTML\u002FCSS patterns exponentially during generation. Running against 34-billion-parameter models, the system achieved 55 to 60 percent token compression, cut compute cost by 44.9 percent, and reduced end-to-end inference time by 48.8 percent. The code is open-source on GitHub.\n\nUI-to-code generation is already a crowded space — tools from Vercel, Builder.io, and various AI startups all promise to turn designs into deployable front-ends. What separates this work is the focus on cost and speed rather than raw output fidelity, which matters when the models doing this work run at 34 billion parameters and the inference bills scale accordingly. A near-50 percent inference speedup is the kind of number that changes whether a feature is economically viable to ship.\n\nThe results are researcher-reported and benchmarked on the team's own test suite, so independent replication will determine how well those numbers hold in production — a gap that tends to be larger than the paper suggests.","[\"ai\",\"dev-tools\",\"open-source\",\"machine-learning\"]","2026-06-30T04:00:00.000Z","2026-06-30T14:26:17.049Z","2026-06-30T14:26:20.148Z","published",null,[],"ai",[24,26,27,28],"dev-tools","open-source","machine-learning",[30],{"name":31,"url":32},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2509.12159",0,{"sections":35},[36,40,45,50,55,60,65,70,75,79,84,88,93,98],{"name":37,"slug":24,"count":38,"latest_published_at":39},"AI",2590,"2026-07-16T04:00:00.000Z",{"name":41,"slug":42,"count":43,"latest_published_at":44},"Security","security",294,"2026-07-15T19:59:48.000Z",{"name":46,"slug":47,"count":48,"latest_published_at":49},"Deals","deals",179,"2026-06-29T20:02:07.000Z",{"name":51,"slug":52,"count":53,"latest_published_at":54},"Policy","policy",158,"2026-07-16T00:02:48.000Z",{"name":56,"slug":57,"count":58,"latest_published_at":59},"Hardware","hardware",122,"2026-07-14T19:46:26.000Z",{"name":61,"slug":62,"count":63,"latest_published_at":64},"Consumer Tech","consumer-tech",93,"2026-07-13T13:20:48.000Z",{"name":66,"slug":67,"count":68,"latest_published_at":69},"Software","software",70,"2026-07-13T19:52:25.000Z",{"name":71,"slug":72,"count":73,"latest_published_at":74},"Science","science",66,"2026-07-10T10:29:37.000Z",{"name":76,"slug":26,"count":77,"latest_published_at":78},"Dev Tools",59,"2026-07-07T04:00:00.000Z",{"name":80,"slug":81,"count":82,"latest_published_at":83},"Gaming","gaming",41,"2026-07-09T04:00:00.000Z",{"name":85,"slug":86,"count":82,"latest_published_at":87},"Startups","startups","2026-06-29T20:55:50.000Z",{"name":89,"slug":90,"count":91,"latest_published_at":92},"General","general",29,"2026-07-10T22:28:58.000Z",{"name":94,"slug":95,"count":96,"latest_published_at":97},"Reviews","reviews",20,"2026-06-24T12:00:01.000Z",{"name":99,"slug":100,"count":101,"latest_published_at":102},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]