{"id":424,"date":"2026-09-20T08:05:32","date_gmt":"2026-09-20T00:05:32","guid":{"rendered":"https:\/\/aidashxp.com\/jev-typesafe-review\/"},"modified":"2026-09-20T08:05:32","modified_gmt":"2026-09-20T00:05:32","slug":"jev-typesafe-review","status":"publish","type":"post","link":"https:\/\/aidashxp.com\/en\/jev-typesafe-review\/","title":{"rendered":"Jev deep review: A non-LLM built by ChatGPT\u2019s creator, replacing hallucination with probability"},"content":{"rendered":"<p class=\"wp-block-paragraph\"><strong>Jev<\/strong> yes <strong>TypeSafe AI<\/strong> A brand-new AI model launched this week, founded by <strong>Diogo Almeida<\/strong>\u2014 he was formerly a researcher at OpenAI and helped build <a href=\"https:\/\/chatgpt.com\" target=\"_blank\" rel=\"nofollow noopener\">ChatGPT<\/a>, and co-invented RLHF (Reinforcement Learning from Human Feedback), the foundational technology underpinning today\u2019s large model era. What makes Jev uniquely special is:<strong>It is not a large language model (LLM)<\/strong>, it does not output text\u2014it outputs probabilities.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">core competencies<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">After leaving OpenAI, Almeida founded TypeSafe AI with a singular mission: to make AI truly \u201cuseful,\u201d not merely fluent in human language. He observed that, over the past four years, the industry has optimized for \u201chuman language,\u201d yet computers actually require a different language\u2014<strong>Structured decisions<\/strong>. Jev was built for exactly this purpose.<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Outputs probabilities instead of text<\/strong>: Jev outputs \u201ccalibrated decisions\u201d\u2014judgments accompanied by confidence scores\u2014not blocks of text<\/li>\n<li><strong>No hallucinations<\/strong>: Because the output space is predefined by the user, the model cannot \u201cfabricate\u201d non-existent answers<\/li>\n<li><strong>Extremely low cost<\/strong>: Output tokens are completely free; input tokens are billed per billion\u2014not per million, as is standard across the industry<\/li>\n<li><strong>Extremely fast<\/strong>: By eliminating the language generation step, inference speed increases dramatically<\/li>\n<li><strong>\u201cSystem One\u201d model<\/strong>: Focuses on intuitive judgment rather than step-by-step reasoning<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">Technically, Jev is trained exclusively on synthetic data using a method Almeida calls \u201cReinforcement Learning from Calibrated Decisions\u201d (RLCD). The company keeps its architecture confidential, and external speculation suggests it may be built atop an open-weight LLM.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">User experience\/limitations<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">Jev is currently offered via API, primarily targeting developers for tasks such as<strong>classification, routing, and security validation<\/strong>\u2014automated tasks of this kind. At launch, demand surged so high that the company\u2019s API briefly became unavailable to users.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Real-world test results are impressive: Vercel engineers replaced OpenAI\u2019s <a href=\"https:\/\/chatgpt.com\" target=\"_blank\" rel=\"nofollow noopener\">ChatGPT<\/a> Luna 5.6 for command safety classification and achieved a speedup of <strong>5x to 18x<\/strong>while also improving accuracy; Bryo AI\u2019s CTO benchmarked it against Google Gemini for email classification\u2014Gemini delivered marginally higher accuracy, but at a cost <strong>10x to 20x higher<\/strong>.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Its limitations are clear-cut:<strong>Jev cannot generate text<\/strong>and is suited only for \u201cmaking judgments,\u201d not \u201cwriting content\u201d; additionally, it partially shifts the \u201challucination\u201d problem to users\u2014when the model returns 50% confidence, you must decide whether to trust it. It functions more like an \u201cintelligent validator\u201d or \u201clow-cost router\u201d for LLMs than a replacement.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">Overall Score<\/h2>\n\n\n\n<figure class=\"wp-block-table\"><table>\n<thead><tr><th>\u7ef4\u5ea6<\/th><th>Score<\/th><th>evaluate<\/th><\/tr><\/thead>\n<tbody>\n<tr><td>functional completeness<\/td><td>7.5 \/ 10<\/td><td>Specialized in decision-making\/classification\u2014not a general-purpose model\u2014with narrow capabilities<\/td><\/tr>\n<tr><td>\u6613\u7528\u6027<\/td><td>7.0 \/ 10<\/td><td>Developer-focused, requiring predefined output structure<\/td><\/tr>\n<tr><td>Cost-effectiveness<\/td><td>9.5 \/ 10<\/td><td>Output is free; input is billed per billion tokens\u2014extremely low cost<\/td><\/tr>\n<tr><td>\u4e2d\u6587\u652f\u6301<\/td><td>8.0 \/ 10<\/td><td>Language-agnostic, fully applicable to Chinese-language scenarios<\/td><\/tr>\n<tr><td>\u8f93\u51fa\u8d28\u91cf<\/td><td>8.0 \/ 10<\/td><td>Calibrated probabilities are reliable and hallucination-free, though slightly outperformed by LLMs on some tasks<\/td><\/tr>\n<\/tbody><\/table><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Overall rating: 8.0\/10<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Jev is a \u201ccounter-trend\u201d product: at a time when large models grow ever larger and more expensive, it trades away language generation to gain speed, cost efficiency, and reliability. For developers needing massive-scale automated judgments, it may be a more pragmatic choice than LLMs. To discover more useful AI tools, visit <a href=\"https:\/\/aidashxp.com\/en\/\">AI Dash<\/a>.<\/p>","protected":false},"excerpt":{"rendered":"<p>Jev \u662f TypeSafe AI \u672c\u5468 [&hellip;]<\/p>\n","protected":false},"author":0,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"default","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","ast-disable-related-posts":"","theme-transparent-header-meta":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"default","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"footnotes":""},"categories":[6],"tags":[],"class_list":["post-424","post","type-post","status-publish","format-standard","hentry","category-ai-coding"],"_links":{"self":[{"href":"https:\/\/aidashxp.com\/en\/wp-json\/wp\/v2\/posts\/424","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/aidashxp.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/aidashxp.com\/en\/wp-json\/wp\/v2\/types\/post"}],"replies":[{"embeddable":true,"href":"https:\/\/aidashxp.com\/en\/wp-json\/wp\/v2\/comments?post=424"}],"version-history":[{"count":0,"href":"https:\/\/aidashxp.com\/en\/wp-json\/wp\/v2\/posts\/424\/revisions"}],"wp:attachment":[{"href":"https:\/\/aidashxp.com\/en\/wp-json\/wp\/v2\/media?parent=424"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/aidashxp.com\/en\/wp-json\/wp\/v2\/categories?post=424"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/aidashxp.com\/en\/wp-json\/wp\/v2\/tags?post=424"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}