{"id":13357,"date":"2026-09-19T13:00:00","date_gmt":"2026-09-19T08:00:00","guid":{"rendered":"https:\/\/pivot.uz\/?p=13357"},"modified":"2026-09-19T12:45:45","modified_gmt":"2026-09-19T07:45:45","slug":"ai-bozorida-yangi-model-jev-nima-uchun-dasturchilar-etiborini-tortmoqda","status":"publish","type":"post","link":"https:\/\/pivot.uz\/uz\/ai-bozorida-yangi-model-jev-nima-uchun-dasturchilar-etiborini-tortmoqda\/","title":{"rendered":"AI bozorida yangi model: Jev nima uchun dasturchilar e\u2019tiborini tortmoqda?"},"content":{"rendered":"\n<p class=\"wp-block-paragraph\">TypeSafe AI kompaniyasi Jev nomli yangi AI modelini taqdim etdi. Model ChatGPT, Claude yoki Gemini kabi javoblarni matn ko\u2018rinishida generatsiya qilmaydi. Uning vazifasi \u2014 dastur uchun kerakli qarorni tez va strukturali shaklda qaytarish.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Model 15-sentabr kuni taqdim etilgan va qisqa vaqt ichida dasturchilar orasida katta qiziqish uyg\u2018otgan. Vercel ma\u2019lumotlariga ko\u2018ra, Jev AI Gateway platformasida ishga tushirilganidan keyin 24 soat ichida pullik jamoalarning qariyb 13 foizida ishlatilgan. Bu platformadagi avvalgi model relizlaridan tezroq adoption ko\u2018rsatkichi bo\u2018lgan.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">Jev nima bilan farq qiladi?<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">Hozirgi AI modellari asosan inson bilan muloqot qilish uchun optimallashtirilgan. Foydalanuvchi savol beradi, model esa matn, kod yoki boshqa kontent bilan javob beradi.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Jev&#8217;ga ma\u2019lumot va oldindan belgilangan variantlar beriladi. U esa ushbu variantlardan keraklisini tanlab, qaror bilan birga uning ehtimolini qaytaradi. Masalan, dasturga kelgan mijoz xabarini olaylik. Oddiy LLM undan foydalanib uzun javob yaratishi mumkin. Jev esa xabarni:<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>\u201cto\u2018lov muammosi\u201d \u2014 87% ehtimol<\/strong> yoki <strong>\u201crefund so\u2018rovi\u201d \u2014 91% ehtimol<\/strong> kabi dastur to\u2018g\u2018ridan-to\u2018g\u2018ri ishlata oladigan strukturali natijaga aylantiradi.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">TypeSafe bu yondashuvni <strong>System One Model<\/strong> deb ataydi. Kompaniya ma\u2019lumotiga ko\u2018ra, Jevning chiqishi oldindan belgilangan strukturaga ega bo\u2018lgani uchun model noto\u2018g\u2018ri formatdagi javob yoki o\u2018zi istagan boshqa turdagi matnni generatsiya qila olmaydi.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Jev matnni ketma-ket tokenma-token generatsiya qilmagani uchun uning ishlash mexanizmi an&#8217;anaviy LLMlardan farq qiladi. TypeSafe parallel sampling arxitekturasi va Reinforcement Learning for Calibrated Decisions (RLCD) deb ataladigan yondashuvdan foydalanishini bildirgan.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Kompaniya ma\u2019lumotiga ko\u2018ra, Jevning input narxi <strong>1 mln token uchun $0.042<\/strong>, output esa bepul. Model taxminan <strong>70\u2013500 millisekund<\/strong> ichida javob berishi mumkin.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">TypeSafe o\u2018z benchmarklarida ayrim vazifalarda Jev an&#8217;anaviy LLM workflowlariga nisbatan yuzlab marta tezroq va arzonroq bo\u2018lishi mumkinligini ko\u2018rsatmoqda. Biroq bu ko\u2018rsatkichlar kompaniyaning o\u2018z benchmarklariga tegishli va ularni barcha AI vazifalariga tatbiq etib bo\u2018lmaydi.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Modelning asosiy maqsadi ChatGPT o\u2018rnini bosish emas. Jev dasturlar ichida juda ko\u2018p marta takrorlanadigan kichik qarorlarni avtomatlashtirish uchun mo\u2018ljallangan. Masalan:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>mijoz murojaatini kategoriyalash;<\/li>\n\n\n\n<li>fraud yoki shubhali tranzaksiyani aniqlash;<\/li>\n\n\n\n<li>AI agent uchun keyingi harakatni tanlash;<\/li>\n\n\n\n<li>minglab ma\u2019lumotlarni saralash;<\/li>\n\n\n\n<li>xavfsizlik tekshiruvlarini avtomatlashtirish;<\/li>\n\n\n\n<li>AI tomonidan yaratilgan kontentni tekshirish;<\/li>\n\n\n\n<li>qaysi so\u2018rovni kuchliroq AI modeliga yuborishni aniqlash.<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">Bu yerda Jevning asosiy g\u2018oyasi shundan iborat: har bir kichik qaror uchun qimmat va sekin chatbotni ishga tushirish shart emas.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">AI agentlar va avtomatlashtirilgan dasturlar soni oshgani sari modellar nafaqat odamlar bilan suhbatlashishi, balki boshqa dasturlar ichida minglab kichik qarorlarni qabul qilishi kerak bo\u2018lmoqda.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Masalan, bitta AI agent bir vazifani bajarish davomida o\u2018nlab yoki yuzlab marta \u201cQaysi yo\u2018lni tanlash kerak?\u201d, \u201cBu xabar qaysi kategoriyaga kiradi?\u201d yoki \u201cBu vaziyatda keyingi qadam nima?\u201d kabi qarorlarni qabul qilishi mumkin. Har bir qaror uchun katta til modelidan foydalanish esa xarajat va kechikishni oshiradi.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">TypeSafe aynan shu qatlamni Jev bilan egallashni maqsad qilmoqda. Kompaniya asoschisi Diogo Almeida bundan oldin OpenAI\u2019da ishlagan va ChatGPT\u2019ni yaratishga olib kelgan instruction-following tadqiqotlarida qatnashgan.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Jev bilan birga TypeSafe AI stealth rejimidan chiqdi va kompaniya <strong>$40 mln miqdorida seed investitsiya<\/strong> jalb qilganini e\u2019lon qildi. Raundni DCVC boshqargan. Model hozircha early access formatida taqdim etilmoqda.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Jevning nomi esa iqtisodchi William Stanley Jevons sharafiga qo\u2018yilgan. Bu yerda kompaniya uning \u201cJevons paradoksi\u201d g\u2018oyasiga ishora qiladi: biror resurs arzonlashgani sari undan foydalanish kamaymasdan, aksincha, oshishi mumkin.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">TypeSafe bu g\u2018oyani AI\u2019ga qo\u2018llamoqda: agar \u201cintellekt\u201dni ishlatish xarajati keskin pasaysa, uni dasturlarning juda ko\u2018p joyiga qo\u2018shish mumkin. Shu sababli Jevning eng katta yangiligi uning ChatGPT\u2019dan aqlliroq ekanida emas. Asosiy g\u2018oya \u2014 <strong>AI\u2019dan foydalanish shaklini o\u2018zgartirish<\/strong>.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>TypeSafe AI kompaniyasi Jev nomli yangi AI modelini taqdim etdi. Model ChatGPT, Claude yoki Gemini kabi javoblarni matn ko\u2018rinishida generatsiya qilmaydi. Uning vazifasi \u2014 dastur uchun kerakli qarorni tez va strukturali shaklda qaytarish. Model 15-sentabr kuni taqdim etilgan va qisqa vaqt ichida dasturchilar orasida katta qiziqish uyg\u2018otgan. Vercel ma\u2019lumotlariga ko\u2018ra, Jev AI Gateway platformasida ishga [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":13355,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"jnews-multi-image_gallery":[],"jnews_single_post":{"format":"standard"},"jnews_primary_category":[],"jnews_override_bookmark_settings":[],"jnews_paywall_metabox":[],"jnews_override_counter":[],"jnews_podcast_option":[],"jnews_podcast_series":[],"footnotes":""},"categories":[115],"tags":[],"jnews-series":[],"class_list":["post-13357","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-yangiliklar"],"_links":{"self":[{"href":"https:\/\/pivot.uz\/uz\/wp-json\/wp\/v2\/posts\/13357","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/pivot.uz\/uz\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/pivot.uz\/uz\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/pivot.uz\/uz\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/pivot.uz\/uz\/wp-json\/wp\/v2\/comments?post=13357"}],"version-history":[{"count":1,"href":"https:\/\/pivot.uz\/uz\/wp-json\/wp\/v2\/posts\/13357\/revisions"}],"predecessor-version":[{"id":13358,"href":"https:\/\/pivot.uz\/uz\/wp-json\/wp\/v2\/posts\/13357\/revisions\/13358"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/pivot.uz\/uz\/wp-json\/wp\/v2\/media\/13355"}],"wp:attachment":[{"href":"https:\/\/pivot.uz\/uz\/wp-json\/wp\/v2\/media?parent=13357"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/pivot.uz\/uz\/wp-json\/wp\/v2\/categories?post=13357"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/pivot.uz\/uz\/wp-json\/wp\/v2\/tags?post=13357"},{"taxonomy":"jnews-series","embeddable":true,"href":"https:\/\/pivot.uz\/uz\/wp-json\/wp\/v2\/jnews-series?post=13357"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}