{"rewrite":{"id":"r_fa1318aa3af23a0161e4ac64","clusterId":"c_52f1004368f36631a81c827f","slug":"ternlight-packs-a-5mb-ai-model-for-on-device-website-search","model":"deepseek-v4-flash","headline":"Ternlight Packs a 5MB AI Model for On-Device Website Search","summary":"Ternlight is a new package that lets developers embed a 5-7MB AI model into websites. The model runs on the visitor's CPU, enabling semantic search without external servers. Two variants are available: base (7MB) and mini (5.5MB). Developer Wen Shu Tang released the package on GitHub and npm, and is seeking contributors.","whyItMatters":"The package makes on-device AI search practical for any website, removing server costs and privacy concerns.","webCardHtml":"\u003cp\u003eThe demo site loads the model on access and performs searches with a few milliseconds delay on a laptop, according to the package\u0026#39;s developer. The model is a distilled version of the small embedding model all-MiniLM-L6-v2, packaged in two sizes: @ternlight/base at 7MB and @ternlight/mini at 5.5MB.\u003c/p\u003e\u003cp\u003eWeb developers can use ternlight to add semantic search or other embedding-based features to their sites. The computation happens entirely on the visitor\u0026#39;s device, meaning no data leaves the browser. The package is available on GitHub and npm under an open license. Wen Shu Tang, the developer, is seeking contributors to improve performance and quality.\u003c/p\u003e","blueskyPost":"Ternlight is a new web package that embeds a 5-7MB AI model into websites, running on the visitor's CPU. It enables semantic search without external servers. Developer Wen Shu Tang released it on GitHub and npm, aiming for improvements. As reported by GIGAZINE.","twitterPost":"Ternlight embeds a 5-7MB AI model into websites, running on the visitor's CPU for semantic search. Two variants available. Developer Wen Shu Tang seeks contributors. From GIGAZINE.","threadsPost":null,"newsletterBlurb":"Ternlight is a new package for embedding lightweight AI models directly into websites, with the model running on the visitor's device. The two variants, at 5.5MB and 7MB, enable semantic search without external servers. Developer Wen Shu Tang has released the package on GitHub and npm and is looking for contributors to improve performance.","attributionJson":"[{\"source\":\"GIGAZINE\",\"url\":\"https://gigazine.net/news/20260707-ternlight-webassembly-embeddings/\",\"title\":\"ternlight allows embedding a mere 5MB AI model into websites for local user operations, enabling AI-powered search functions and more\"}]","lintFlagsJson":null,"lintHits":0,"costUsd":0,"inputTokens":4190,"outputTokens":2663,"status":"published","repairAttempts":0,"nextRepairAt":null,"factsAttemptedAt":1783436192,"createdAt":"2026-07-07T14:43:40.000Z","publishedAt":"2026-07-07T14:48:32.000Z","updatedAt":"2026-07-07T14:43:40.000Z"},"cluster":{"id":"c_52f1004368f36631a81c827f","canonicalTitle":"わずか5MBのAIモデルをウェブサイトに組み込んでユーザーにローカル操作させられる「ternlight」が登場、ウェブサイトにAIを活用した検索機能などを追加可能","representativeArticleId":"a_433b5e32991ff91c097696b1","sourceCount":1,"writtenSourceCount":1,"writeAttempts":0,"isSolo":true,"entitiesJson":"{\"anime_titles\":[],\"manga_titles\":[],\"work_titles\":[],\"studios\":[],\"people\":[],\"type\":\"announcement\",\"domain\":\"other\",\"is_roundup\":false}","contentType":"news","status":"published","firstSeenAt":"2026-07-07T14:00:00.000Z","lastSeenAt":"2026-07-07T14:00:00.000Z","updatedAt":"2026-07-07T14:48:32.000Z"},"attribution":[{"source":"GIGAZINE","url":"https://gigazine.net/news/20260707-ternlight-webassembly-embeddings/","title":"わずか5MBのAIモデルをウェブサイトに組み込んでユーザーにローカル操作させられる「ternlight」が登場、ウェブサイトにAIを活用した検索機能などを追加可能"}],"entities":{"anime_titles":[],"manga_titles":[],"work_titles":[],"studios":[],"people":[],"type":"announcement","domain":"other","is_roundup":false},"keyFacts":null}