{"id":111086,"date":"2025-11-18T18:17:46","date_gmt":"2025-11-18T18:17:46","guid":{"rendered":"https:\/\/ekamu.net\/?p=111086"},"modified":"2025-11-18T18:17:46","modified_gmt":"2025-11-18T18:17:46","slug":"google-en-gelismis-yapay-zeka-modeli-gemini-3u-tanitti","status":"publish","type":"post","link":"https:\/\/ekamu.net\/index.php\/2025\/11\/18\/google-en-gelismis-yapay-zeka-modeli-gemini-3u-tanitti\/","title":{"rendered":"Google, en geli\u015fmi\u015f yapay zeka modeli Gemini 3\u2019\u00fc tan\u0131tt\u0131"},"content":{"rendered":"

Google, bug\u00fcnden itibaren Gemini 3<\/strong> ad\u0131n\u0131 verdi\u011fi yeni yapay zeka modellerini kullan\u0131c\u0131larla bulu\u015fturmaya ba\u015flad\u0131. \u015eirket, bu seriyi \u201cen ak\u0131ll\u0131<\/strong>\u201d ve ger\u00e7eklere dayal\u0131 do\u011frulu\u011fu y\u00fcksek AI sistemleri olarak tan\u0131ml\u0131yor. Gemini 3, \u00f6zellikle OpenAI\u2019\u0131n GPT-5 lansman\u0131ndaki sorunlar\u0131n ard\u0131ndan Google\u2019\u0131n t\u00fcketici odakl\u0131 yapay zekada \u00f6ne ge\u00e7mesini sa\u011flayacak \u00f6nemli bir ad\u0131m olarak g\u00f6r\u00fcl\u00fcyor. Demis Hassabis ve Koray Kavuk\u00e7uo\u011flu \u00f6nc\u00fcl\u00fc\u011f\u00fcndeki DeepMind ekibi, Gemini 3\u2019\u00fc \u201cher fikri hayata ge\u00e7irebilecek en zeki modelimiz\u201d olarak nitelendiriyor.<\/p>\n

Gemini 2.5\u2019in \u00fczerinden sadece yedi ay ge\u00e7mi\u015fken tan\u0131t\u0131lan bu yeni s\u00fcr\u00fcm, Google\u2019\u0131n bug\u00fcne kadarki en yetenekli b\u00fcy\u00fck dil modeli (LLM) olarak \u00f6ne \u00e7\u0131k\u0131yor. Yeni serinin en dikkat \u00e7ekici \u00fcyesi olan Gemini 3 Pro<\/strong>, tan\u0131t\u0131m g\u00fcn\u00fcnden itibaren Gemini uygulamas\u0131<\/strong> \u00fczerinden t\u00fcm kullan\u0131c\u0131lara a\u00e7\u0131lm\u0131\u015f durumda. Ayn\u0131 zamanda ABD\u2019deki Google AI Pro ve Ultra aboneleri, Gemini 3 Pro\u2019yu AI Modu ile Google Arama i\u00e7inde deneyimleyebilecek.<\/p>\n

Her alanda ciddi geli\u015fme var<\/b><\/p>\n

Gemini 3 Pro, do\u011fal olarak \u00e7ok modlu bir model olarak tasarland\u0131. Yani metin, g\u00f6rsel ve ses verilerini ayn\u0131 anda i\u015fleyebiliyor. \u00d6rne\u011fin, bir tarifin foto\u011fraflar\u0131n\u0131 al\u0131p bunu bir yemek kitab\u0131na d\u00f6n\u00fc\u015ft\u00fcrebiliyor veya bir video ders serisinden interaktif kartlar olu\u015fturabiliyor.<\/p>\n

Yeni modelin iyile\u015ftirmeleri Google \u00fcr\u00fcnlerinin \u00e7o\u011funda hissedilecek. Gemini uygulamas\u0131nda kullan\u0131c\u0131lar, daha kapsaml\u0131 programlar olu\u015fturabilecekleri bir \u00e7al\u0131\u015fma alan\u0131na kavu\u015fuyor. Ayr\u0131ca Gemini Labs\u2019ta test edilen \u00fcretken aray\u00fczler sayesinde, Gemini 3 Pro g\u00f6rsel ve dergi tarz\u0131 i\u00e7erikler veya \u00f6zel kullan\u0131c\u0131 aray\u00fczleri tasarlayabiliyor. AI Modu ile Arama\u2019da ise g\u00f6rseller, tablolar, \u0131zgaralar ve sim\u00fclasyonlar gibi g\u00f6rsel \u00f6gelerle sorgular\u0131n\u0131za yan\u0131t sunuluyor. Google, g\u00fcncellenmi\u015f \u201csorgu par\u00e7alama\u201d tekni\u011fi sayesinde, modelin sorular\u0131n niyetini daha iyi anlayarak \u00f6nceki aramalarda g\u00f6zden ka\u00e7m\u0131\u015f i\u00e7erikleri bulabildi\u011fini vurguluyor.<\/p>\n

Google, Gemini 3 Pro\u2019yu ChatGPT\u2019nin bo\u015f \u00f6vg\u00fclerine g\u00f6re daha g\u00fcvenilir bir model olarak tan\u0131ml\u0131yor. Google DeepMind \u00dcr\u00fcn Ba\u015fkan\u0131 Tulsee Doshi, yan\u0131tlar\u0131n art\u0131k ak\u0131ll\u0131, \u00f6z ve do\u011frudan<\/strong> oldu\u011funu, kli\u015fe ve a\u015f\u0131r\u0131 iltifat yerine ger\u00e7ek bilgiler sundu\u011funu ifade ediyor. Yani model art\u0131k \u00e7ok daha az yalakal\u0131k<\/strong> yap\u0131yor.<\/p>\n

Gemini 3 Pro\u2019nun mant\u0131k y\u00fcr\u00fctme ve g\u00f6rev tamamlama yetenekleri de geli\u015ftirildi. Daha karma\u015f\u0131k g\u00f6revleri yerine getirebiliyor <\/strong>ve uzun vadeli planlamalar yapabiliyor<\/strong>. Bu \u00f6zellikler, kullan\u0131c\u0131 ad\u0131na e-posta inceleme, seyahat ara\u015ft\u0131rmas\u0131 ve rezervasyon gibi i\u015flemleri ger\u00e7ekle\u015ftirebilen Gemini Agent <\/strong>deneysel \u00f6zelli\u011finde kullan\u0131l\u0131yor.<\/p>\n

Ayr\u0131ca, yaln\u0131zca test kullan\u0131c\u0131lar\u0131na a\u00e7\u0131k olan Deep Think modu<\/strong>, mant\u0131k y\u00fcr\u00fctme kapasitesini daha da art\u0131r\u0131yor. Deepthink, g\u00fcvenlik testlerinden ge\u00e7tikten sonra \u00f6n\u00fcm\u00fczdeki haftalarda Google AI Ultra<\/strong> abonelerine sunulacak.<\/p>\n

Gemini 3, geli\u015ftiriciler i\u00e7in de g\u00fc\u00e7l\u00fc ara\u00e7lar sunuyor. S\u0131f\u0131rdan \u00fcretim yetene\u011fiyle karma\u015f\u0131k talimatlar\u0131 i\u015fleyerek daha zengin ve etkile\u015fimli web aray\u00fczleri olu\u015fturabiliyor. Agentic kodlama ve vibe kodlama alan\u0131nda en yetkin model olarak WebDev Arena\u2019da 1487 Elo puan\u0131 al\u0131yor. Terminal-Bench 2.0 ve SWE-bench Verified testlerinde 2.5 Pro\u2019ya k\u0131yasla \u00f6nemli \u00fcst\u00fcnl\u00fck sa\u011fl\u0131yor. Model, Google AI Studio, Vertex AI, Gemini CLI ve yeni agentic geli\u015ftirme platformu Google Antigravity<\/strong> ile kullan\u0131labiliyor. Ayr\u0131ca Cursor, GitHub, JetBrains, Manus ve Replit gibi \u00fc\u00e7\u00fcnc\u00fc taraf platformlarda da eri\u015filebilir.<\/p>\n

Gemini 3 ile e\u015f zamanl\u0131 olarak tan\u0131t\u0131lan Antigravity<\/strong> adl\u0131 kodlama aray\u00fcz\u00fc, \u00e7oklu panelli ve ajan destekli bir deneyim sunuyor. Warp veya Cursor 2.0 gibi<\/strong> ara\u00e7lara benzer \u015fekilde \u00e7al\u0131\u015fan Antigravity, ChatGPT tarz\u0131 bir komut penceresini, bir terminal aray\u00fcz\u00fcn\u00fc ve de\u011fi\u015fikliklerin etkilerini g\u00f6steren bir taray\u0131c\u0131 penceresini bir araya getiriyor.<\/p>\n

\u00dcst\u00fcn mant\u0131k ve \u00e7ok modlu yetkinlik<\/b><\/p>\n

<\/span> Performans taraf\u0131na bakt\u0131\u011f\u0131m\u0131zda model, ba\u011f\u0131ms\u0131z \u00f6l\u00e7\u00fcmlerde de g\u00fc\u00e7l\u00fc performans g\u00f6steriyor. Humanity\u2019s Last Exam testinde elde etti\u011fi 37.4 puan ile t\u00fcm zamanlar\u0131n en y\u00fcksek skorunu yakalad\u0131. Ayr\u0131ca kullan\u0131c\u0131 memnuniyetini \u00f6l\u00e7en LMArena s\u0131ralamas\u0131nda da zirvede yer al\u0131yor. Matematik alan\u0131nda MathArena Apex\u2019te y\u00fczde 23,4 ile yeni bir standart belirledi. \u00c7ok modlu becerilerinde de \u00f6ne \u00e7\u0131kan Gemini 3 Pro, MMMU-Pro\u2019da y\u00fczde 81, Video-MMMU\u2019da y\u00fczde 87,6 ve SimpleQA Verified\u2019de y\u00fczde 72,1 ba\u015far\u0131yla karma\u015f\u0131k problemlere g\u00fcvenilir \u00e7\u00f6z\u00fcmler sunuyor. <\/figure>\n
<\/span> Deep Think modu ise modelin mant\u0131k ve \u00e7ok modlu anlay\u0131\u015f kapasitesini daha da ileri ta\u015f\u0131yor. Testlerde, Gemini 3 Deep Think, Humanity\u2019s Last Exam\u2019de y\u00fczde 41 ve GPQA Diamond\u2019da y\u00fczde 93,8 skorlar\u0131yla Gemini 3 Pro\u2019yu geride b\u0131rak\u0131yor. Ayr\u0131ca ARC-AGI-2\u2019de y\u00fczde 45,1 ile yeni ve benzersiz zorluklar\u0131 \u00e7\u00f6zme yetene\u011fini g\u00f6steriyor. <\/figure>\n

Gemini 3 d\u00f6nemi ba\u015flad\u0131<\/b><\/p>\n

Google\u2019a g\u00f6re Gemini uygulamas\u0131n\u0131n ayl\u0131k 650 milyon aktif kullan\u0131c\u0131s\u0131 bulunuyor ve 13 milyon yaz\u0131l\u0131m geli\u015ftirici modeli i\u015f ak\u0131\u015flar\u0131nda aktif olarak kullan\u0131yor.<\/p>\n

Gemini 3\u2019\u00fcn da\u011f\u0131t\u0131m\u0131 bug\u00fcn itibar\u0131yla<\/strong> geni\u015f bir ekosisteme yay\u0131lm\u0131\u015f durumda. Model, Gemini uygulamas\u0131nda t\u00fcm kullan\u0131c\u0131lara a\u00e7\u0131l\u0131rken, AI Mode \u00fczerinden Google AI Pro ve Ultra aboneleri de yeni yeteneklerden yararlanabiliyor. Geli\u015ftiriciler, Gemini 3\u2019e Gemini API, AI Studio, Google Antigravity ve Gemini CLI arac\u0131l\u0131\u011f\u0131yla eri\u015febiliyor. Kurumsal tarafta ise model, Vertex AI ve Gemini Enterprise platformlar\u0131na entegre edildi. Daha geli\u015fmi\u015f mant\u0131k modunu sunan Gemini 3 Deep Think ise ek g\u00fcvenlik de\u011ferlendirmelerinin ard\u0131ndan \u00f6n\u00fcm\u00fczdeki haftalarda Google AI Ultra abonelerine ula\u015facak.<\/p>\n\n

https:\/\/www.youtube.com\/embed\/98DcoXwGX6I<\/span><\/p>\n

Kaynak :\u00a0https:\/\/www.donanimhaber.com\/google-en-gelismis-yapay-zeka-modeli-gemini-3-u-tanitti–198724<\/span><\/p>\n","protected":false},"excerpt":{"rendered":"

Google, bug\u00fcnden itibaren Gemini 3 ad\u0131n\u0131 verdi\u011fi yeni yapay zeka modellerini kullan\u0131c\u0131larla bulu\u015fturmaya ba\u015flad\u0131. \u015eirket, bu seriyi \u201cen ak\u0131ll\u0131\u201d ve ger\u00e7eklere dayal\u0131 do\u011frulu\u011fu y\u00fcksek AI sistemleri olarak tan\u0131ml\u0131yor. Gemini 3, \u00f6zellikle OpenAI\u2019\u0131n GPT-5 …<\/p>\n","protected":false},"author":1,"featured_media":111087,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[8],"tags":[1324,6591,321,233,452],"class_list":["post-111086","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-teknoloji","tag-gemini","tag-gemini-3","tag-google","tag-kullanici","tag-model"],"_links":{"self":[{"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/posts\/111086","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/comments?post=111086"}],"version-history":[{"count":1,"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/posts\/111086\/revisions"}],"predecessor-version":[{"id":111090,"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/posts\/111086\/revisions\/111090"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/media\/111087"}],"wp:attachment":[{"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/media?parent=111086"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/categories?post=111086"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/ekamu.net\/index.php\/wp-json\/wp\/v2\/tags?post=111086"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}