{"id":66854,"date":"2024-07-19T09:24:06","date_gmt":"2024-07-19T09:24:06","guid":{"rendered":"https:\/\/kamucalisani.net\/?p=66854"},"modified":"2024-07-19T09:24:06","modified_gmt":"2024-07-19T09:24:06","slug":"openai-yeni-ucuz-yapay-zeka-modeli-gpt-4o-miniyi-tanitti","status":"publish","type":"post","link":"https:\/\/kamucalisani.net\/index.php\/2024\/07\/19\/openai-yeni-ucuz-yapay-zeka-modeli-gpt-4o-miniyi-tanitti\/","title":{"rendered":"OpenAI, yeni ucuz yapay zeka modeli GPT-4o Mini’yi tan\u0131tt\u0131"},"content":{"rendered":"

<\/span> OpenAI, GPT-4o Mini <\/strong>isimli daha hafif ve daha ucuz<\/strong> yapay zeka modelini tan\u0131tt\u0131. Tam boyutlu modellerden \u00e7ok daha d\u00fc\u015f\u00fck maliyetli olan modelin GPT-3.5’ten daha yetenekli <\/strong>oldu\u011fu belirtiliyor. <\/figure>\n

OpenAI’\u0131n modellerini kullanarak uygulamalar geli\u015ftirmek isteyen ancak maliyeti nedeniyle kar\u015f\u0131layamayan geli\u015ftiriciler i\u00e7in geli\u015ftirilen model, Google’\u0131n Gemini 1.5 Flash<\/strong>‘\u0131 ve Anthropic’in Claude 3 Haiku<\/strong>‘su gibi mini modellere rakip olacak. OpenAI, GPT-4o Mini’nin geli\u015ftiriciler i\u00e7in milyon jeton (token) girdi ba\u015f\u0131na 15 sent<\/strong> ve milyon jeton \u00e7\u0131kt\u0131 ba\u015f\u0131na 60 sent<\/strong> fiyatland\u0131rmaya sahip oldu\u011funu ve GPT-3.5 Turbo’dan %60’tan daha ucuz <\/strong>oldu\u011funu belirtiyor.<\/p>\n

GPT-3.5’in yerini alacak<\/b><\/p>\n

Bug\u00fcnden itibaren \u00fccretsiz, Plus ve Team planlar\u0131na sahip ChatGPT kullan\u0131c\u0131lar\u0131, GPT-3.5 Turbo yerine GPT-4o Mini’yi kullanabilecek. Kurumsal kullan\u0131c\u0131lar i\u00e7in ise \u00f6n\u00fcm\u00fczdeki hafta eri\u015fime a\u00e7\u0131lacak. GPT-4o Mini, bundan b\u00f6yle GPT-3.5’in yerini alacak.<\/strong> Ancak, geli\u015ftiriciler GPT-4o Mini’ye ge\u00e7meyi tercih etmez ise, API arac\u0131l\u0131\u011f\u0131yla GPT-3.5’i \u015fimdilik kullanabilecek.<\/p>\n

Yeni model ayn\u0131 zamanda API’de metin ve g\u00f6r\u00fcnt\u00fc deste\u011fine<\/strong> de sahip olacak. OpenAI, modelin yak\u0131nda video ve ses gibi t\u00fcm \u00e7ok modlu girdi ve \u00e7\u0131kt\u0131lar\u0131 i\u015fleyebilece\u011fini <\/strong>s\u00f6yl\u00fcyor.\u00a0Bunun yan\u0131nda model,128 bin<\/strong> jeton\u00a0ba\u011flam penceresinin yan\u0131nda, istek ba\u015f\u0131na 16 bin<\/strong> jetona kadar \u00e7\u0131kt\u0131 destekliyor ve Ekim 2023’e kadar <\/strong>bilgilere cevap verebiliyor.<\/p>\n

OpenAI, GPT-4o Mini’nin rakiplerinden daha iyi s\u00f6yl\u00fcyor<\/b><\/p>\n

<\/span> GPT-4o Mini, 57 akademik konuda yakla\u015f\u0131k 16.000 \u00e7oktan se\u00e7meli sorudan olu\u015fan bir k\u0131yaslama s\u0131nav\u0131 olan Measuring Massive Multitask Language Understanding’de (MMLU) 100 \u00fczerinden 82 puan<\/strong> elde etti. Bu testte GPT-3.5 70 puan<\/strong> al\u0131rken, GPT-4o 88,7 puan<\/strong> al\u0131yor. Google Gemini Ultra ise 90 puanla<\/strong> zirvede yer al\u0131yor. GPT-4o Mini’ye\u00a0rakip modellere bakt\u0131\u011f\u0131m\u0131zda, Claude 3 Haiku ve Gemini 1.5 Flash’\u0131n s\u0131ras\u0131yla 75,2 ve 78,9 puan <\/strong>ald\u0131\u011f\u0131n\u0131 g\u00f6r\u00fcyoruz. OpenAI,\u00a0 GPT-4o Mini’nin di\u011fer bir\u00e7ok kar\u015f\u0131la\u015ft\u0131rma testinde rakiplerinden daha iyi performans g\u00f6sterdi\u011fini iddia ediyor.<\/strong> <\/figure>\n

GPT-4o Mini’nin kullan\u0131ma sunulmas\u0131yla, ucuza yapay zeka uygulamalar\u0131 olu\u015fturmak isteyen geli\u015ftiriciler yeni ve yetenekli bir alternatife kavu\u015fmu\u015f oluyor.<\/p>\n\n

Kaynak\u00a0 :\u00a0https:\/\/www.donanimhaber.com\/openai-yeni-ucuz-yapay-zeka-modeli-gpt-4o-mini-yi-tanitti–179628<\/span><\/p>\n","protected":false},"excerpt":{"rendered":"

OpenAI, GPT-4o Mini isimli daha hafif ve daha ucuz yapay zeka modelini tan\u0131tt\u0131. Tam boyutlu modellerden \u00e7ok daha d\u00fc\u015f\u00fck maliyetli olan modelin GPT-3.5’ten daha yetenekli oldu\u011fu belirtiliyor. OpenAI’\u0131n modellerini kullanarak uygulamalar …<\/p>\n","protected":false},"author":1,"featured_media":66855,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[8],"tags":[3568,538,3567,449,159],"class_list":["post-66854","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-teknoloji","tag-gpt-3-5","tag-gpt-4o","tag-mini","tag-model","tag-openai"],"_links":{"self":[{"href":"https:\/\/kamucalisani.net\/index.php\/wp-json\/wp\/v2\/posts\/66854","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/kamucalisani.net\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/kamucalisani.net\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/kamucalisani.net\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/kamucalisani.net\/index.php\/wp-json\/wp\/v2\/comments?post=66854"}],"version-history":[{"count":1,"href":"https:\/\/kamucalisani.net\/index.php\/wp-json\/wp\/v2\/posts\/66854\/revisions"}],"predecessor-version":[{"id":66858,"href":"https:\/\/kamucalisani.net\/index.php\/wp-json\/wp\/v2\/posts\/66854\/revisions\/66858"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/kamucalisani.net\/index.php\/wp-json\/wp\/v2\/media\/66855"}],"wp:attachment":[{"href":"https:\/\/kamucalisani.net\/index.php\/wp-json\/wp\/v2\/media?parent=66854"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/kamucalisani.net\/index.php\/wp-json\/wp\/v2\/categories?post=66854"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/kamucalisani.net\/index.php\/wp-json\/wp\/v2\/tags?post=66854"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}