{"rewrite":{"id":"r_5d95991c8021a9667dbcd7b3","clusterId":"c_d03518e97e5026b08b3840d5","slug":"jev-router-selects-ai-models-by-task-difficulty-on-openrouter","model":"deepseek-v4-1-flash","headline":"Jev Router Selects AI Models By Task Difficulty On OpenRouter","summary":"TypeSafe AI's Jev Router became available on OpenRouter, the AI gateway service, on September 28, 2026. The router sends a user's prompt through Jev, the judgment model TypeSafe AI released on September 15, 2026, to score the difficulty and accuracy of the task, then assigns an appropriate AI model. Easy tasks go to inexpensive small models and difficult tasks to more expensive state-of-the-art models, cutting the waste that comes from routing easy and difficult work to the same model. OpenRouter's earlier routers analyze the type of task but not its difficulty. Testing across four AI agent benchmarks found Jev Router completed 82% more tasks than OpenRouter's Auto Router. Jev Router does not save prompts and does not use them for training. TypeSafe AI is led by CEO Diogo Almeida, who contributed to the development of ChatGPT.","whyItMatters":"Routing has been priced on task type alone, so a prompt that needed a large model and one that did not cost the same; Jev Router prices the decision itself and reports 82% more benchmark tasks completed than OpenRouter's Auto Router.","webCardHtml":"\u003cp\u003e\u003c/p\u003e\u003cul\u003e\n\u003cli\u003eJev is described as a \u0026#34;System One model,\u0026#34; named for the fast, intuitive judgment Nobel Prize-winning economist Daniel Kahneman called System 1, distinct from slower System 2 deliberation.\u003c/li\u003e\n\u003cli\u003eTypeSafe AI describes Jev as \u0026#34;a new class of state-of-the-art model built to make rapid, structured decisions that software can use directly.\u0026#34;\u003c/li\u003e\n\u003cli\u003eJev uses a training method TypeSafe AI calls \u0026#34;Reinforcement Learning from Correct Decisions (RLCD),\u0026#34; and returns only the decision rather than text.\u003c/li\u003e\n\u003cli\u003eASCII.jp reports Jev costs $0.042 per million input tokens, with output free. A roughly two-minute test battle using about 3 million tokens cost $0.126, about 19 yen.\u003c/li\u003e\n\u003cli\u003eNew accounts get $5 worth of tokens free, which ASCII.jp\u0026#39;s tester had not used up after extensive use.\u003c/li\u003e\n\u003cli\u003eHTML implementing Jev for game NPC control and debugging, including a battle where the player character sent two kinds of input every 0.2 seconds.\u003c/li\u003e\n\u003cli\u003eGIGAZINE links Jev to earlier coverage describing it as capable of \u0026#34;GPT-5.6 Terra-class performance\u0026#34; cheaply and at ultra-high speed.\u003c/li\u003e\n\u003cli\u003eA separate compact decision model, \u0026#34;Jeff,\u0026#34; runs locally at about 22 milliseconds per decision on an RTX PRO 6000 and about 28 milliseconds on an Apple M4 Max.\u003c/li\u003e\n\u003c/ul\u003e\u003cp\u003e\u003c/p\u003e","blueskyPost":"Jev Router's privacy claim is separate from its accuracy claim. Prompts are not saved and not used for training, which matters for teams that cannot send client work to a model whose logs they cannot inspect.","twitterPost":"Jev Router scores difficulty with Jev, a judgment model from TypeSafe AI. The untold part: OpenRouter gains a routing layer it did not build.","threadsPost":"Jev Router beats OpenRouter's Auto Router on four agent benchmarks because it reads difficulty, not task type. It runs on OpenRouter itself, so the gateway is now shipping a routing decision made by TypeSafe AI. OpenRouter's own router still does not score difficulty.","newsletterBlurb":"OpenRouter now offers Jev Router, which runs a prompt through TypeSafe AI's Jev model to score difficulty before assigning a model. Easy work goes to small cheap models, hard work to state-of-the-art ones. Across four AI agent benchmarks it completed 82% more tasks than Auto Router.","attributionJson":"[{\"source\":\"GIGAZINE\",\"url\":\"https://gigazine.net/news/20260928-openrouter-jev-router/\",\"title\":\"Jev Router, which automatically selects which AI to use via Jev, launches; it can instantly determine the appropriate AI model from task difficulty with higher accuracy than OpenRouter's existing routers\"},{\"source\":\"ASCII.jp\",\"url\":\"https://ascii.jp/elem/000/004/437/4437496/?rss\",\"title\":\"AI Plays Games Like a Human: What I Learned from Trying the Buzzworthy \\\"Jev\\\"\"},{\"source\":\"GIGAZINE\",\"url\":\"https://gigazine.net/news/20260929-jeff/\",\"title\":\"Free Jev-compatible compact high-speed decision model \\\"Jeff\\\" runs locally at blazing speed of about 22-28 milliseconds\"}]","lintFlagsJson":null,"lintHits":0,"costUsd":0,"inputTokens":13698,"outputTokens":1268,"status":"published","repairAttempts":0,"nextRepairAt":null,"factsAttemptedAt":1791044968,"createdAt":"2026-10-03T16:23:30.000Z","publishedAt":"2026-10-03T16:28:20.000Z","updatedAt":"2026-10-03T16:28:20.000Z"},"cluster":{"id":"c_d03518e97e5026b08b3840d5","canonicalTitle":"どのAIを使うかをJevで自動選択する「Jev Router」が登場、タスク難易度から適切なAIモデルを瞬時に判断可能でOpenRouterの既存ルーターより高精度","representativeArticleId":"a_4e309933fa289029b9178edf","sourceCount":2,"writtenSourceCount":2,"writeAttempts":0,"isSolo":false,"entitiesJson":"{\"anime_titles\":[],\"manga_titles\":[],\"work_titles\":[\"Jev\",\"Jev Router\",\"Jeff\"],\"studios\":[\"TypeSafe AI\"],\"people\":[\"Diogo Almeida\"],\"type\":\"news\",\"domain\":\"other\",\"is_roundup\":false}","contentType":"news","status":"published","firstSeenAt":"2026-09-28T04:20:00.000Z","lastSeenAt":"2026-09-30T08:52:00.000Z","updatedAt":"2026-10-03T17:11:51.000Z"},"attribution":[{"source":"ASCII.jp","url":"https://ascii.jp/elem/000/004/437/4437496/?rss","title":"AIが人間のようにゲームをプレイ　話題の「Jev」を試してわかったこと"},{"source":"GIGAZINE","url":"https://gigazine.net/news/20260930-d1-decision-model-liquid-ai/","title":"ベンチマークでJev超えな意思決定モデル「d1」が登場"}],"entities":{"anime_titles":[],"manga_titles":[],"work_titles":["Jev","Jev Router","Jeff"],"studios":["TypeSafe AI"],"people":["Diogo Almeida"],"type":"news","domain":"other","is_roundup":false},"keyFacts":["Jev Router became available on OpenRouter on September 28, 2026.","Jev Router completed 82% more tasks than OpenRouter's Auto Router across four AI agent benchmarks.","Jev was released by TypeSafe AI on September 15, 2026, and returns decisions with confidence scores.","TypeSafe AI prices Jev at $0.042 per million input tokens, with output free.","TypeSafe AI is led by CEO Diogo Almeida, a co-developer of ChatGPT."]}