Ziyosfera

Gemini (til modeli): Versiyalar orasidagi farq

yangi
#YOʻNALTIRISH [[Gemini (til modeli)]]
{{Dastur bilgiqutisi
| logo = [[File:Google Gemini logo 2025.svg|frameless|upright=1.15]]
| author =
| developer = [[Google AI]]<br>[[Google DeepMind]]
| released = {{Start date and age|2023|12|06}} (beta version)<br>{{Start date and age|2024|02|08}} (official rollout)
| discontinued =
| latest release version = 3.1 Pro, 3 Deep Think, 3.8 Flash, 3.5 Flash-Lite <ref name="G3Flash" /><ref name="G3.1Pro" />
| latest release date = {{Start date and age|2026|09|02}} (3.8 Flash)
| qid =
| programming language =
| replaces = [[PaLM]]
| replaced_by =
| language = English and 70+ other languages
}}
 
'''Gemini''' — [[Google DeepMind]] tomonidan ishlab chiqilgan [[multimodal learning|multimodal]] [[large language model|katta til modellari]] (LLM) oilasi boʻlib, [[LaMDA]] va [[PaLM|PaLM 2]] modellarining vorisi hisoblanadi. U Gemini Pro, Gemini Deep Think, Gemini Flash va Gemini Flash Lite modellaridan iborat.<ref name="G3Flash" /> Model 2023-yil 6-dekabrda eʼlon qilingan. Gemini [[Gemini (chatbot)|Gemini]] [[chatbot|suhbatdoshi]]ga (chatbot) asos boʻladi. Uning nomi [[Gemini (astrology)|Egizaklar]] [[Astrological sign|burji]] sharafiga qoʻyilgan.
 
== Tarixi ==
=== Ishlab chiqilishi ===
 
[[Google]] kompaniyasi oʻzining shoʻʼba korxonasi [[Google DeepMind]] ishlab chiqqan [[large language model|katta til modeli]] Geminini 2023-yil 10-mayda [[Google I/O]] konferensiyasining asosiy maʼruzasida eʼlon qildi. Gemini shu tadbirda taqdim etilgan [[PaLM 2]] modelining yanada qudratli vorisi sifatida koʻrsatildi. Google [[Chief executive officer|bosh direktori]] [[Sundar Pichai]] Gemini hali ishlab chiqishning dastlabki bosqichida ekanini aytdi.<ref name="IO2023NYT" /><ref name="IO2023ZD" /> Boshqa katta til modellaridan farqli oʻlaroq, Gemini faqat [[text corpus|matn korpusi]] asosida oʻqitilmagan. U [[Multimodal learning|multimodal]] boʻlishi uchun loyihalangan, yaʼni bir vaqtning oʻzida matn, tasvir, audio, video va [[Source code|dastur kodi]] kabi turli xil maʼlumot turlarini qayta ishlay oladi.<ref name="AnnounceGuardian" /> Model Google DeepMind tarkibiga birlashtirilgan Google kompaniyasining ikki boʻlimi — DeepMind va [[Google Brain]] hamkorligida yaratilgan.<ref name="Google25" /> ''[[Wired (magazine)|Wired]]'' nashriga bergan intervyusida DeepMind bosh direktori [[Demis Hassabis]] Gemininig ilgʻor imkoniyatlarini taʼrifladi. Uning fikricha, bu imkoniyatlar algoritmga [[OpenAI]] kompaniyasining [[GPT-4]] asosida ishlaydigan [[ChatGPT]] xizmatidan ustun kelish imkonini beradi. ChatGPTning ortib borayotgan mashhurligiga Google [[LaMDA]] va [[Bard (chatbot)|Bard]] bilan qattiq raqobat qilib kelayotgan edi. Hassabis DeepMindning [[AlphaGo]] dasturi yutuqlarini alohida taʼkidladi. Bu dastur 2016-yilda [[Go (game)|Go]] chempioni [[Lee Sedol|Li Sedol]]ni magʻlub etib, butun dunyo eʼtiborini tortgandi. Uning soʻzlariga koʻra, Gemini AlphaGo va Google–DeepMindning boshqa katta til modellari qudratini oʻzida birlashtiradi.<ref name="HassabisInterview" />
 
2023-yil avgustda ''[[The Information (website)|The Information]]'' nashri Googlening Gemini boʻyicha rejalari haqida maʼruza eʼlon qildi. Unda kompaniya modelni 2023-yil oxirida ishga tushirishni moʻljallagani aytilgandi. Maʼruzaga koʻra, Google koʻpchilik katta til modellariga xos suhbat imkoniyatlarini [[artificial intelligence|sunʼiy intellekt]] yordamida tasvir yaratish bilan birlashtirib, OpenAI va boshqa raqobatchilardan oʻzib ketishga umid qilgan. Bu esa modelga kontekstga mos tasvirlar yaratish va koʻproq [[use cases|qoʻllash holatlari]]ga moslashish imkonini bergan boʻlardi.<ref name="TheInformation" /> Bard singari,<ref name="LarrySergey" /> Google asoschilaridan biri [[Sergey Brin]] ham nafaqadan Gemini ustida ishlash uchun qaytarildi. U bilan birga Google Brain va DeepMinddan yuzlab muhandislar jalb qilindi;<ref name="TheInformation" /><ref name="BrinReturn" /> keyinchalik Brin Gemininig «asosiy hissa qoʻshuvchisi» sifatida qayd etildi.<ref name="CoreContributor" /> Gemini [[YouTube]] videolarining matn transkriptlari asosida oʻqitilgani bois, mualliflik huquqi bilan himoyalangan materiallarni saralab tashlash uchun huquqshunoslar jalb etildi.<ref name="TheInformation" />
 
Gemini yaqinda ishga tushishi haqidagi xabarlardan soʻng, OpenAI GPT-4 ni Geminiga oʻxshash multimodal imkoniyatlar bilan birlashtirish ishlarini tezlashtirdi.<ref name="Hustle" /> ''The Information'' sentabr oyida bir necha kompaniyaga katta til modelining «erta versiyasi»dan foydalanishga ruxsat berilgani haqida xabar berdi. Google bu modelni mijozlarga [[Google Cloud Platform|Google Cloud]]ning Vertex AI xizmati orqali taqdim etishni rejalashtirgandi. Nashr yana Google Geminini ham GPT-4, ham [[Microsoft]] kompaniyasining [[GitHub Copilot]] xizmatiga raqib qilib tayyorlayotganini yozdi.<ref name="NearLaunch" /><ref name="WhatWeKnow" />
 
=== Ishga tushirilishi ===
2023-yil 6-dekabrda Pichai va Hassabis onlayn matbuot anjumanida «Gemini 1.0» ni eʼlon qildi.<ref name="AnnounceWSJ" /><ref name="AnnounceAP" /> U uchta modeldan iborat: «juda murakkab vazifalar» uchun moʻljallangan Gemini Ultra; «keng koʻlamli vazifalar» uchun moʻljallangan Gemini Pro; «qurilma ichidagi vazifalar» uchun moʻljallangan Gemini Nano. Ishga tushirilgan paytda Gemini Pro Bard tarkibiga, Gemini Nano esa [[Pixel 8 Pro]] smartfoniga integratsiya qilingandi. Gemini Ultra «Bard Advanced»ga asos boʻlishi va 2024-yil boshida dasturchilar uchun ochilishi rejalashtirilgandi. Google Geminini yana [[Google Search|Search]], [[Google Ads|Ads]], [[Google Chrome|Chrome]], [[Google Workspace]]dagi Duet AI va [[AlphaCode 2]] mahsulotlariga joriy etmoqchi edi.<ref name="AnnounceArs" /><ref name="AnnounceAP" /> Model faqat ingliz tilida taqdim etildi.<ref name="AnnounceAP" /><ref name="AnnounceVerge" /> Kompaniya Geminini oʻzining «eng yirik va eng qudratli sunʼiy intellekt modeli» deb taʼrifladi hamda uni inson xatti-harakatlarini taqlid qiladigan qilib loyihalaganini aytdi.<ref name="AnnounceCNN" /><ref name="AnnounceAP" /><ref name="AnnounceCBS" /> Shu bilan birga, «keng qamrovli xavfsizlik sinovlari» zarurligi sababli model keyingi yilgacha ommaga taqdim etilmasligi maʼlum qilindi.<ref name="AnnounceWSJ" /> Gemini Googlening [[Tensor Processing Unit|Tensor Processing Unit]] (TPU) protsessorlarida oʻqitilgan va shular asosida ishlaydi.<ref name="AnnounceWSJ" /><ref name="AnnounceVerge" /> Modelning nomi DeepMind va Google Brain birlashuviga, shuningdek, [[NASA]]ning [[Project Gemini|Gemini loyihasi]]ga ishoradir.<ref name="AnnounceWired" />
 
Gemini Ultra bir qator soha koʻrsatkichlarida GPT-4, [[Anthropic]] kompaniyasining [[Claude 2]], [[Inflection AI]] kompaniyasining Inflection-2, [[Meta Platforms|Meta]] kompaniyasining [[LLaMA 2]] va [[xAI (company)|xAI]] kompaniyasining [[Grok 1]] modellaridan ustun kelgani aytildi.<ref name="AnnounceTime" /><ref name="AnnounceWSJ" /> Gemini Pro esa [[GPT-3.5]]dan ustun chiqqani qayd etildi.<ref name="AnnounceGuardian" /> Bundan tashqari, Gemini Ultra 57 ta fanni qamrab oluvchi [[Massive Multitask Language Understanding]] (MMLU) testida inson mutaxassislaridan ustun kelgan birinchi til modeli boʻldi va 90% natija koʻrsatdi.<ref name="AnnounceGuardian" /><ref name="AnnounceWired" /> Gemini Pro 13-dekabrda Google Cloud mijozlariga AI Studio va Vertex AI orqali taqdim etildi. Gemini Nano esa [[Android (operating system)|Android]] dasturchilari uchun ham ochiladi.<ref name="AnnounceNYT" /><ref name="AnnounceCNBC" /><ref name="AnnounceBloomberg" /> Hassabis yana DeepMind Gemini «dunyo bilan jismonan oʻzaro taʼsirlashish uchun robototexnika bilan qanday birlashtirilishi mumkinligini» oʻrganayotganini maʼlum qildi.<ref name="NewBreed" /> AQSh prezidenti [[Joe Biden|Jo Bayden]] oktabr oyida imzolagan [[Executive Order 14110|farmon]]ga muvofiq, Google Gemini Ultra sinov natijalarini [[federal government of the United States|AQSh federal hukumati]] bilan boʻlishishini bildirdi. Shu tariqa kompaniya noyabr oyida [[Bletchley Park|Bletchli Park]]da oʻtkazilgan [[AI Safety Summit 2023|Sunʼiy intellekt xavfsizligi sammiti]]da belgilangan tamoyillarga rioya qilish yuzasidan [[government of the United Kingdom|Buyuk Britaniya hukumati]] bilan muzokaralar olib bordi.<ref name="AnnounceGuardian" />
 
2025-yil iyun oyida Google Gemini CLI dasturini taqdim etdi. Bu — Gemini imkoniyatlarini bevosita terminalga olib keluvchi ochiq kodli sunʼiy intellekt agenti. U ilgʻor kod yozish, avtomatlashtirish va muammo yechish vositalarini taklif etadi hamda yakka dasturchilar uchun keng bepul foydalanish chegaralarini beradi.<ref name="kf0Qv" />
 
=== Yangilanishlar ===
2025-yil yanvar oyida Google [[Samsung Electronics|Samsung]] bilan hamkorlik qilib, Gemini Nano va Gemini Proni [[Galaxy S24]] smartfonlari qatoriga joriy etdi.<ref name="S24Bloomberg" /><ref name="S24Wired" /> Keyingi oyda Bard va Duet AI Gemini brendi ostida birlashtirildi.<ref name="GeminiRebrandNYT" /><ref name="GeminiRebrandReuters" /> «Gemini Advanced with Ultra 1.0» esa [[Google One]] obuna xizmatining yangi «AI Premium» tarifi orqali chiqarildi.<ref name="AIPremium" /> Gemini Pro ham global miqyosda ishga tushirildi.<ref name="Global" />
 
2024-yil fevral oyida Google Gemini 1.5 modelini taqdim etib, uni 1.0 Ultradan ham qudratli va imkoniyatliroq model deb taʼrifladi.<ref name="1.5Wired" /><ref name="1.5Forbes" /><ref name="1.5FC" /> Oʻzgarishlar orasida yangi arxitektura, [[mixture-of-experts|mutaxassislar aralashmasi]] yondashuvi va bir million tokenli kattaroq [[context window|kontekst oynasi]] bor.<ref name="1.5BI" /> Shu oyda Google Gemma modellarini taqdim etdi — bu Gemini modellarining kichikroq, [[free and open-source|erkin va ochiq kodli]] turkumi. Bir qancha nashrlar buni Meta va boshqa kompaniyalarning sunʼiy intellekt modellarini ochiq kodga oʻtkazishiga javob deb baholadi. Ular buni Googlening sunʼiy intellektni yopiq saqlash boʻyicha oldingi amaliyotidan voz kechishi sifatida ham izohladi.<ref name="GemmaFC" /><ref name="GemmaBloomberg" /><ref name="GemmaNYT" /> Google 14-mayda 2024-yilgi I/O asosiy maʼruzasida yana bir model — Gemini 1.5 Flashni eʼlon qildi.<ref name="Flash" /> Shu tadbirda Google Gemini Nanoning ish stoli uchun moslashtirilgan versiyasini «Built-in AI» arxitekturasi orqali bevosita Google Chrome brauzeriga joriy etish rejasini maʼlum qildi. Bu Prompt API («window.ai») kabi tajriba bosqichidagi API vositalari orqali veb-dasturchilarga mahalliy hisoblash imkoniyatlarini ochib beradi.
 
Gemininig ikkita yangilangan modeli — Gemini-1.5-Pro-002 va Gemini-1.5-Flash-002 2024-yil 24-sentabrda chiqarildi.<ref name="c65D6" />{{primary inline|date=February 2026}}
 
2024-yil 11-dekabrda Google yangi Gemini 2.0 Flash Experimental modelini eʼlon qildi.<ref name="SLeze" /> Uning imkoniyatlari orasida real vaqtdagi audio va video muloqot uchun Multimodal Live API, ichki tasvir yaratish va boshqariladigan matndan nutq hosil qilish (raqamli belgi bilan), shuningdek, integratsiyalangan Google Search bor.<ref name="S4jSs" /> Model yana takomillashgan agentlik imkoniyatlarini, yangi Google Gen AI SDK toʻplamini va GitHub uchun tajriba bosqichidagi «Jules» nomli sunʼiy intellekt kod yozish agentini taqdim etadi.<ref name="UEHpr" />{{primary inline|date=February 2026}}
 
2025-yil 30-yanvarda Google Gemini 2.0 Flashni yangi standart model sifatida chiqardi, Gemini 1.5 Flash esa foydalanish uchun ochiq qoldi. Shundan soʻng 2025-yil 5-fevralda Gemini 2.0 Pro taqdim etildi. Bundan tashqari, Google Gemini 2.0 Flash Thinking Experimental modelini chiqardi. Bu model soʻrovlarga javob berayotganda til modelining fikrlash jarayoni haqida qisqacha maʼlumot hosil qiladi.<ref name="MOz3i" />{{primary inline|date=February 2026}}
 
2025-yil 12-martda Google [[Gemini Robotics]] modelini ham eʼlon qildi. Bu Gemini 2.0 modellari oilasiga asoslangan [[vision-language-action model|koʻrish-til-harakat modeli]].<ref name="h3GOR" />{{primary inline|date=February 2026}} Ertasi kuni Google [[Android Studio]]dagi Gemini oddiy interfeys [[mockup|maketlari]]ni tushunib, ularni ishlaydigan [[Jetpack Compose]] kodiga aylantira olishini maʼlum qildi.<ref name="8KRAo" />
 
Gemini 2.5 Pro Experimental 2025-yil 25-martda chiqarildi.<ref name="GoogleBlog25" /><ref name="TechRadar25" /><ref name="RDWorld25" /> Bu model tugʻma [[multimodal interaction|multimodallik]]ni saqlagan holda [[chain-of-thought prompting|fikrlash zanjiri]] imkoniyatini taklif etadi.<ref name="GoogleBlog25" /><ref name="RDWorld25" /><ref name="AnalyticsVidhya25" />
 
Google I/O 2025 tadbirida Google Gemini 2.5 Flash standart model boʻlishini eʼlon qildi.<ref name="googleblog" /> Gemini 2.5 Pro esa eng ilgʻor Gemini modeli sifatida taqdim etildi. U yaxshilangan kod yozish imkoniyatlari va boshqa takomillashuvlarga ega.<ref name="k9kC2" /> Ham 2.5 Pro, ham 2.5 Flash tugʻma audio chiqishni qoʻllab-quvvatlaydi.
 
2025-yil 17-iyunda Google 2.5 Pro va 2.5 Flashning umumiy foydalanishga ochilganini eʼlon qildi. Shu kuni ular Gemini 2.5 Flash-Lite modelini ham taqdim etdi — bu tezlik va narx samaradorligi uchun moslashtirilgan model.<ref name="Doshi-2025" />{{primary inline|date=February 2026}}
 
2025-yil 18-noyabrda Google 3 Pro va 3 Deep Think modellarining chiqqanini eʼlon qildi.<ref name="Gemini_3_Pro_DeepThink_Announcement" /> Bu yangi modellar 2.5 Pro va 2.5 Flash oʻrnini egalladi. Mazkur chiqarilish OpenAI kompaniyasini raqib model GPT-5.2 ni tezroq chiqarishga undadi; u 11-dekabrda taqdim etildi.<ref name="pzb1P" /><ref name="8bOyM" /><ref name="twyqf" /><ref name="0wlt5" />
 
2025-yil 4-dekabrda Google 3 Deep Think Ultra obunachilariga bosqichma-bosqich yetkazila boshlashini eʼlon qildi.<ref name="a7HKh" />{{primary inline|date=February 2026}}
 
2025-yil 17-dekabrda Google 3 Flash modeli chiqqanini eʼlon qildi; u 2.5 Flashning joriy versiyasi oʻrnini egalladi.<ref name="9QWuW" />{{primary inline|date=February 2026}}
 
2026-yil 12-yanvarda Apple Siri yordamchisining kelgusi versiyasida Gemini sunʼiy intellekt modelidan foydalanish rejasini eʼlon qildi.<ref name="1wUTT" />
 
2026-yil 11-fevralda Gemini ayollar moda brendi [[L'agence|L'Agence]] bilan hamkorlik qildi: [[New York Fashion Week|Nyu-York moda haftaligi]]da kompaniyaning virtual kiyib koʻrish texnologiyasidan foydalanildi.<ref>{{Cite web |title=Fashion’s biggest brands leaning into AI |url=https://www.nbcnews.com/video/fashion-s-biggest-brands-leaning-into-ai-257682501710 |access-date=2026-04-27 |website=NBC News |language=en}}</ref>
 
2026-yil 19-fevralda Google Gemini 3.1 Pro modelini chiqardi.<ref name="G3.1Pro" /><ref name="G3.1Pro-sec" />
 
2026-yil 26-fevralda Nano Banana 2 taqdim etildi. Bu Gemini 3.1 Flash Image platformasi asosida qurilgan yangilangan versiya.<ref name="zOJDk" />
 
2026-yil 3-martda Google Gemini 3.1 Flash Lite modelini Google API orqali dasturchilarga taqdim etdi.<ref name="v0a1t" /><ref name="vAd8z" />
 
2026-yil 2-aprelda Google Gemma 4 modelini chiqardi. U ilgʻor mulohaza yuritish va agentlik ish jarayonlari uchun maxsus yaratilgan. <ref name="WmcRt" />
 
2026-yil 19-mayda Google Gemini 3.5 Flash modelini taqdim etdi — bu 3.5 modellari oilasidagi birinchi ommaviy nashr. <ref>{{Cite web |date=2026-05-19 |title=Gemini 3.5: frontier intelligence with action |url=https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-5/ |access-date=2026-06-17 |website=Google |language=en-us}}</ref>
 
2026-yil 21-iyulda Google Gemini 3.6 Flash va Gemini 3.5 Flash-Lite modellarini chiqardi.<ref>{{Cite web |date=2026-07-21 |title=Introducing Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber |url=https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-6-flash-3-5-flash-lite-3-5-flash-cyber/ |access-date=2026-07-21 |website=Google |language=en-us}}</ref><ref>{{Cite web |last=Sigalos |first=MacKenzie |date=2026-07-21 |title=Google expands Gemini lineup with cheaper models and new Mythos rival |url=https://www.cnbc.com/2026/07/21/google-gemini-flash-ai-mythos-rival.html |access-date=2026-07-21 |website=CNBC |language=en}}</ref>
[[File:Top 20 OpenRouter LLMs.webp|thumb|Eng yaxshi 20 ta [[OpenRouter]] katta til modeli]]
2026-yil 13-avgustda Google Gemini 3.7 Flash modelini chiqardi.<ref>{{Cite web |date=2026-08-13 |title=Introducing Gemini 3.7 Flash |url=https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-3-7-flash/ |access-date=2026-08-13 |website=Google |language=en-us}}</ref>
 
2026-yil 2-sentabrda Google Gemini 3.8 Flash va 3.8 Flash Cyber modellarini chiqardi.<ref>{{Cite web |date=2026-09-02 |title=Introducing Gemini 3.8 Flash and 3.8 Flash Cyber |url=https://blog.google/innovation-and-ai/models-and-research/gemini-models/3-8-flash-and-3-8-flash-cyber/ |access-date=2026-09-06 |website=Google |language=en-us}}</ref>
 
{{-}}
=== Model versiyalari ===
{{See also|Gemma (language model)}}
Quyidagi jadvalda Gemininig asosiy model versiyalari keltirilgan hamda har bir versiyadagi muhim oʻzgarishlar tavsiflangan:<ref name="mLYmE" /><ref name="4xunj" /><ref>{{Cite web |title=Gemini deprecations  {{!}}  Gemini API  {{!}}  Google AI for Developers |url=https://ai.google.dev/gemini-api/docs/deprecations |archive-url=https://web.archive.org/web/20260526122242/https://ai.google.dev/gemini-api/docs/deprecations |archive-date=2026-05-26 |access-date=2026-05-27 |website=Google AI for Developers |language=en| url-status=live}}</ref>
 
{| class="wikitable sortable"
! Version
! Release date
! Status<ref name="dDYsD" /><ref name="Doshi-2025" />
! Description
|-
| 1.0 Nano
| {{date|2023-12-06}}
| {{version|o|Discontinued}}
| Designed for on-device tasks and first available in Google's [[Pixel 8 Pro]].<ref name="gemini1" />
|-
| 1.0 Pro
| {{date|2023-12-13}}
| {{version|o|Discontinued}}
| Designed for a diverse range of tasks.<ref name="gemini1" />
|-
| 1.0 Ultra
| {{date|2024-02-08}}
| {{version|o|Discontinued}}
| Google's most powerful offering in the Gemini 1.0 family.<ref name="gemini1" />
|-
| 1.5 Pro
| {{date|2024-02-15}}
| {{version|o|Discontinued}}
|As a successor to the 1.0 series of models, 1.5 Pro offers significantly increased context size (up to 1 million tokens). It is designed to be the most capable model in the Gemini 1.5 family.<ref name="gemini1.5" />
|-
| 1.5 Flash
| {{date|2024-05-14}}
|{{version|o|Discontinued}}
|This version got renamed from 'Nano' to 'Flash'. It is also Gemini's free model.
|-
| 2.0 Flash
| {{date|2025-01-30}}
| {{version|o|Discontinued}}
| Developed by Google with a focus on multimodality, agentic capabilities, and speed.<ref name="ZRrBE" />
|-
| 2.0 Flash-Lite
| {{date|2025-02-25}}
| {{version|o|Discontinued}}
| First-ever Gemini Flash-Lite model designed for cost-efficiency and speed.<ref name="w5kMd" />
|-
|2.0 Pro
| {{date|2025-02-05}}
| {{version|o|Discontinued}}
|
|-
| 2.5 Pro
| {{date| 2025-03-25}}
| {{version|c|Latest version}}
|
|-
| 2.5 Flash
| {{date| 2025-04-17}}
| {{version|co|Supported}}
|An incremental improvement from Gemini 2.5.
|-
| 2.5 Flash-Lite
| {{date| 2025-06-17}}
| {{version|co|Supported}}
|
|-
| 2.5 Flash Image ([[Nano Banana]])
| {{date| 2025-08-26}} (Publicly)
| {{version|co|Supported}}
|
|-
|3 Pro
|{{date|2025-11-18}}
|{{version|o|Discontinued preview}}<ref name="OSdTC" />
|Gemini 3 Pro is not a modification or a fine-tune or a prior model. It is the foundational model of the Gemini 3 series. A [[Sparse MoE|sparse mixture-of-experts]], natively multimodal, reasoning model. It can process text, visual, and audio inputs. It supports up to 1,000,000 input tokens and 64,000 output tokens.<ref name="t5REb" />
|-
|3 Pro Image (Nano Banana Pro)
| {{date|2025-11-20}}
| {{version|c|Latest version}}
|An improved version of Nano Banana, which includes better text rendering and better real world knowledge. It was released as a preview model on {{date|2025-11-20}}, and made generally available on {{date|2026-05-28}}.<ref name="x46Kx" />
|-
|3 Deep Think
|{{date|2025-12-03}}
|{{version|p|Preview}}
|Based on the 2.5 Pro "Deep Think" mode that achieved high ratings in IOI<ref name="wWwmU" /><ref name="S5d8b" />
|-
| 3 Flash
| {{date|2025-12-17}}
| {{version|p|Preview}}
|Based on Gemini 3 Pro.
|-
| 3.1 Pro
| {{date|2026-02-19}}
| {{version|p|Preview}}
|Based on Gemini 3 Pro. Created to handle complex reasoning tasks across large datasets and diverse inputs, including text, audio, image, video, and entire codebases.<ref>{{Cite web |title=Gemini 3.1 Pro - Model Card |url=https://deepmind.google/models/model-cards/gemini-3-1-pro/ |access-date=2026-07-23 |website=Google DeepMind |language=en}}</ref>
|-
| 3.1 Flash Image
(Nano Banana 2)
| {{date|2026-02-26}}
| {{version|c|Latest version}}
|Based on Gemini 3 Flash. Designed to generate both image and text outputs.<ref>{{Cite web |title=Gemini 3.1 Flash Image - Model Card |url=https://deepmind.google/models/model-cards/gemini-3-1-flash-image/ |access-date=2026-07-23 |website=Google DeepMind |language=en}}</ref>
|-
|3.1 Flash-Lite
| {{date|2026-03-03}}
| {{version|co|Supported}}
|Based on Gemini 3 Pro. It is optimized for high-volume, latency-sensitive tasks such as translation, and text classification.<ref>{{Cite web |title=Gemini 3.1 Flash-Lite - Model Card |url=https://deepmind.google/models/model-cards/gemini-3-1-flash-lite/ |access-date=2026-07-23 |website=Google DeepMind |language=en}}</ref>
|-
|3.5 Flash
| {{date|2026-05-19}}
| {{version|co|Supported}}
|Based on Gemini 3 Flash.<ref>{{Cite web |last=Bellan |first=Rebecca |date=2026-05-19 |title=With Gemini 3.5 Flash, Google bets its next AI wave on agents, not chatbots |url=https://techcrunch.com/2026/05/19/with-gemini-3-5-flash-google-bets-its-next-ai-wave-on-agents-not-chatbots/ |access-date=2026-05-19 |website=TechCrunch |language=en-US}}</ref>
|-
|3.1 Flash-Lite Image
(Nano Banana 2 Lite)
| {{date|2026-06-30}}
| {{version|c|Latest version}}
|Based on Gemini 3.1 Flash-Lite. It is optimized for cost-effective image generation.<ref>{{Cite web |title=Gemini 3.1 Flash-Lite Image - Model Card |url=https://deepmind.google/models/model-cards/gemini-3-1-flash-lite-image/ |access-date=2026-07-30 |website=Google DeepMind |language=en}}</ref>
|-
|3.5 Flash-Lite
| {{date|2026-07-21}}
| {{version|c|Latest version}}
|Based on Gemini 3.1 Flash-Lite. It is optimized for high-volume, latency-sensitive tasks such as translation, text classification, and supporting agentic workflows.<ref>{{Cite web |title=Gemini 3.5 Flash-Lite - Model Card |url=https://deepmind.google/models/model-cards/gemini-3-5-flash-lite/ |access-date=2026-07-23 |website=Google DeepMind |language=en}}</ref>
|-
|3.6 Flash
| {{date|2026-07-21}}
| {{version|co|Supported}}
|Based on Gemini 3.5 Flash. Offers improvements in coding, general knowledge, and multimodal processing, alongside increased token efficiency relative to its predecessor.<ref>{{Cite web |title=Gemini 3.6 Flash - Model Card |url=https://deepmind.google/models/model-cards/gemini-3-6-flash/ |access-date=2026-07-23 |website=Google DeepMind |language=en}}</ref>
|-
|3.7 Flash
| {{date|2026-08-13}}
| {{version|co|Supported}}
|Based on Gemini 3.6 Flash. It features improvements to its core reasoning foundation.<ref>{{Cite web |title=Gemini 3.7 Flash - Model Card |url=https://deepmind.google/models/model-cards/gemini-3-7-flash/ |access-date=2026-08-13 |website=Google DeepMind |language=en}}</ref>
|-
|3.8 Flash
| {{date|2026-09-02}}
| {{version|c|Latest version}}
|Based on Gemini 3.7 Flash.<ref>{{Cite web |title=Gemini 3.8 Flash - Model Card |url=https://deepmind.google/models/model-cards/gemini-3-8-flash/ |access-date=2026-09-02 |website=Google DeepMind |language=en}}</ref> A version with a more permissive set of mitigations for cybersecurity, 3.8 Flash Cyber, is available through Google's Fairwind Program.
|}
 
== Nano Banana ==
{{redirect-distinguish|Google Pics|Google Photos|Google Images}}
{{also|AI content watermarking}}
[[File:ChaoticHarmony explosion in post-painter abstraction (Nano Banana 2).webp|alt=AI-generated abstract art created with Nano Banana 2 at Gemini|thumb|Geminida Nano Banana 2 yordamida yaratilgan sunʼiy intellekt [[abstract art]]]]
'''Nano Banana''' (rasmiy nomi '''Gemini 2.5 Flash Image'''), '''Nano Banana Pro''' (rasmiy nomi '''Gemini 3 Pro Image''') va '''Nano Banana 2''' (rasmiy nomi '''Gemini 3.1 Flash Image''') — [[Artificial intelligence visual art|tasvir yaratish]] va [[Image editing|tahrirlash]] modellari.
 
«Nano Banana» — bu modelning [[Arena (AI platform)|Arena]] platformasida yashirin ommaviy sinovdan oʻtayotgan paytdagi kod nomi edi. U birinchi marta 2025-yil 12-avgustda Arena nomli sunʼiy intellekt baholash platformasida anonim model sifatida paydo boʻldi. Model 2025-yil 26-avgustda Gemini ilovasi va [[Google AI]] xizmatlari orqali ommaga taqdim etildi. «Nano Banana» laqabi [[Google DeepMind]] mahsulot menejeri Naina Raisinghaniga berilgan laqablardan kelib chiqqan.<ref name="kGaLa" /> Keyinchalik Google rasmiy eʼlonida uning asl nomi ''Gemini 2.5 Flash Image'' ekanini tasdiqladi.<ref name="elpais2" /><ref name="gblog2" /> 2025-yil 20-noyabrda DeepMind Nano Banana Pro (Gemini 3 Pro Image) modelini chiqardi. Unda matn tasvirlash va dunyo haqidagi bilim sezilarli yaxshilandi.<ref name="x46Kx" /><ref name="vv1gn" /><ref name="DtCVU" /><ref name="gblog2" />
 
Chiqqanidan soʻng Nano Banana ijtimoiy tarmoqlarda [[Viral phenomenon|virusli internet hodisasiga]] aylandi. Ayniqsa, uning [[photorealistic|fotorealistik]] «[[3D figurine|3D haykalcha]]» tasvirlari katta shuhrat qozondi. Model Gemini ilovasi, Google AI Studio va Vertex AI orqali foydalanuvchilarga taqdim etildi. Googlening maʼlum qilishicha, u Gemini ilovasiga 10 milliondan ortiq yangi foydalanuvchi jalb qilgan. Shuningdek, ishga tushirilganidan bir necha hafta ichida 200 milliondan ziyod tasvir tahriri amalga oshirilgan.<ref name="dQrC4" /><ref name="dd94J" />
 
Model foydalanuvchilarga soch turmagi va fonni oʻzgartirish hamda tabiiy tildagi koʻrsatmalar yordamida suratlarni birlashtirish imkonini beradi. Obyekt izchilligi tufayli bir xil shaxs yoki narsa tasvirning turli tahrirlarida ham tanib olinadi. Koʻp tasvirni birlashtirish funksiyasi bir necha fotosuratni yaxlit bir natijaga aylantiradi, dunyo haqidagi bilim esa kontekstga mos oʻzgarishlar kiritishga yordam beradi. Model, shuningdek, SynthID [[Digital watermarking|raqamli belgisini]] qoʻyadi. Bu — natijalardagi koʻrinmas raqamli imzo boʻlib, sunʼiy intellekt yaratgan maʼlumotni aniqlashga xizmat qiladi.<ref name="gblog2" /><ref name="pcgamer2" />  Koʻp tasvirni birlashtirish funksiyasi bir necha fotosuratni yaxlit bir natijaga aylantiradi, dunyo haqidagi bilim esa kontekstga mos oʻzgarishlar kiritishga yordam beradi. Odamlar Nano Bananani virusli mashhur boʻlgan bir moda bilan bogʻlay boshladi: unda foydalanuvchilar oʻz selfilarini oʻyinchoqqa oʻxshash 3D haykalchalarga aylantirgan. Bu hodisa [[Instagram]] va [[Twitter|X (avvalgi Twitter)]] kabi saytlarda tez tarqaldi.<ref name="auto32" /><ref name="HHNth" /> Model X tarmogʻiga qoʻshilgach, foydalanuvchilar postlarda Nano Bananani bevosita belgilab, koʻrsatmalar asosida surat yarata olishdi. Bu esa uning shuhratini yanada oshirdi.<ref name="auto32" />
 
''[[TechRadar]]''ning 2025-yil sentabrdagi sharhida Nano Banana turli koʻrsatmalarda [[GPT Image|ChatGPTning tasvir yaratish]] xizmatiga qaraganda realroq va izchilroq ekani qayd etildi.<ref name="v1ey9" /> ''[[Tom's Guide]]'' sharhida uning ijodiy va jonli tasvir tahrirlarini bajara olishi yuqori baholandi.<ref name="JFRoK" /> ''[[PC Gamer]]'' sharhida esa modelda [[Cropping (image)|kesish]] kabi ayrim oddiy tahrir vositalari yoʻqligi taʼkidlandi. Bundan tashqari, mahsulot baʼzan oʻzgarishlarni qoʻllamay, asl tasvirni qaytargani ham qayd etildi.<ref name="pcgamer2" /> Nano Banana [[architectural visualization|meʼmoriy vizualizatsiya]] sohasida yaxshi natija koʻrsatdi: u murakkab geometriyada ham tasvirlarni toʻgʻri miqyosda yaratdi.<ref name="lLFT0" /><ref name="pcgamer2" />
 
2026-yil 26-fevralda Nano Banana 2 ishga tushirildi hamda Gemini chatboti, Search AI Mode va Lens xizmatlariga integratsiya qilindi. Bu Gemini 3.1 Flash Image asosidagi tezroq versiya boʻlib, koʻrsatmalarga amal qilish va matn tasvirlash imkoniyatlari yaxshilangan.<ref name="zOJDk" /> May oyida boʻlib oʻtgan 2026-yilgi [[Google I/O]] konferensiyasida Google Nano Banana asosidagi '''Google Pics''' nomli tasvir yaratish platformasini eʼlon qildi. U [[Google Workspace]] toʻplamining bir qismi hisoblanadi.<ref>{{Cite web |last=Peters|first=Jay |date=May 19, 2026 |title=Google Pics is a new app that tries to fix AI image editing |url=https://www.theverge.com/tech/933026/google-pics-app-workspace-ai-images-io-2026 |url-status=live |archive-url=https://web.archive.org/web/20260519182501/https://www.theverge.com/tech/933026/google-pics-app-workspace-ai-images-io-2026 |archive-date=May 19, 2026 |access-date=May 19, 2026 |website=[[The Verge]]}}</ref>
 
2026-yil 30-iyulda Google Nano Bananani [[Google Earth]] bilan integratsiya qildi. Ammo bu xizmat 24 soat ichida oʻchirildi. Sababi — tadqiqotchilar, ochiq manba tekshiruvchilari va ijtimoiy tarmoq foydalanuvchilarining noroziligi boʻldi, chunki xizmat notoʻgʻri maʼlumot tarqatishga moyil edi. Foydalanuvchilar koʻrsatmalarda cheklovlar yoʻqligini isbotlash uchun yirik falokatlar va harbiy harakatlarning soxta tasvirlarini yaratdi. Ular orasida eng koʻzga tashlangani [[September 11th attacks|11-sentabr terror hujumlari]] manzarasini qayta yaratish boʻldi.<ref>{{cite news |last1=Gibbs |first1=Alice |title=Google Earth pulls “Nano Banana” after users create offensive images |url=https://www.newsweek.com/google-earth-pulls-nano-banana-after-users-create-offensive-images-12276889 |access-date=3 August 2026 |work=Newsweek |date=3 August 2026 |language=en}}</ref><ref>{{cite news |last1=Hassan |first1=Adeel |title=Google Earth Disables A.I. Tool After One Day Over Disinformation Concerns |url=https://www.nytimes.com/2026/08/02/technology/google-earth-ai-satellite-images.html |access-date=3 August 2026 |work=The New York Times |date=2 August 2026}}</ref><ref>{{cite news |last1=Wong |first1=Matteo |title=Google Just Ruined One of Its Most Important Tools |url=https://www.theatlantic.com/technology/2026/07/google-earth-ai-images/688145/ |access-date=3 August 2026 |work=The Atlantic |date=31 July 2026 |language=en}}</ref>
 
== Texnik tavsiflar ==
 
Gemini multimodal boʻlgani uchun har bir kontekst oynasi bir necha xil kirish maʼlumotini oʻz ichiga olishi mumkin. Turli rejimlarni aralashtirib qoʻllash mumkin, ularni qatʼiy tartibda berish shart emas. Bu multimodal suhbat oʻtkazishga imkon beradi. Masalan, foydalanuvchi suhbatni matn, rasm, video va audio aralashmasi bilan istalgan tartibda boshlashi mumkin. Gemini ham xuddi shunday erkin tartibda javob berishi mumkin. Kiritiladigan tasvirlar turli [[Image resolution|oʻlchamlarda]] boʻlishi mumkin, video esa tasvirlar ketma-ketligi sifatida kiritiladi. Audio 16 [[kHz]] chastotada namunalanadi, soʻngra Universal Speech Model uni tokenlar ketma-ketligiga aylantiradi. Gemininig maʼlumotlar toʻplami multimodal va koʻp tilli boʻlib, u «veb-hujjatlar, kitoblar va kod hamda tasvir, audio va video maʼlumotlarni» oʻz ichiga oladi.<ref name="WhitePaper" />
 
Gemini va [[Gemma (language model)|Gemma]] modellari faqat dekoderdan iborat [[Transformer (deep learning architecture)|transformerlardir]]. Ularga TPU qurilmalarida samarali oʻqitish va xulosa chiqarish uchun oʻzgartirishlar kiritilgan. 1.0 avlodi [[Transformer (deep learning architecture)#Multi-Query Attention|koʻp soʻrovli eʼtibor]] mexanizmidan foydalanadi.<ref name="WhitePaper" /> Biroq Gemini 2.0, 2.5 va 3 uchun oq kitoblar chop etilmagan.
 
{| class="wikitable sortable"
|+Technical specifications of Gemini models
! Generation
! Variant
! Release date
! Parameters
! Context length
! Notes
|-
| rowspan=4|1.0
| Nano-1
| {{date|2023-12-06}}
| 1.8B
| rowspan=4|32,768
| rowspan=2|[[Knowledge distillation|Distilled]] from "larger Gemini models", 4-bit quantized<ref name="WhitePaper" />
|-
| Nano-2
| {{date|2023-12-06}}
| 3.25B
|-
| Pro
| {{date|2023-12-13}}
| Not publicly known
|
|-
| Ultra
| {{date|2024-02-08}}
| Not publicly known
|
|-
| rowspan=2| 1.5
| Pro
| {{date|2024-02-15}}
| Not publicly known
| rowspan=2| 10,000,000<ref name="q3e3Z" /><ref name="WhitePaper1.5" />
| rowspan=2|1 million tokens in production API
|-
| Mini
| {{date|2024-05-14}}
|Not publicly known
|}
 
== Baholanishi ==
{{See also|Google Gemini#Reception}}
 
''[[MIT Technology Review]]'' Gemini taqdimoti atrofidagi mish-mishlarni «sunʼiy intellekt shov-shuvining choʻqqisi» deb taʼrifladi.<ref name="PeakHype" /><ref name="AnnounceTime" /> ''[[Business Insider]]''dan Hugh Langley Gemini Google uchun hal qiluvchi lahza ekanini yozdi: «Agar Gemini hayratga solsa, bu Googlega Microsoft va OpenAI uni ogohlantirmay ortda qoldirdi degan fikrni oʻzgartirishga yordam beradi. Agar u umidni oqlamasa, Google ortda qoldi deydigan tanqidchilarni yanada dadillashtiradi».<ref name="Hsiao" />
 
2023-yil dekabrdagi taqdimotga munosabat bildirgan informatika olimi [[Oren Etzioni]] [[Google DeepMind|Google Deepmind]] va [[OpenAI]] oʻrtasida [[arms race|qurollanish poygasi]] boshlanishini bashorat qildi. Informatika olimi [[Alexei A. Efros|Alexei Efros]] Gemininig multimodal yondashuvi salohiyatini yuqori baholadi.<ref name="AnnounceWired" /> Professorlar [[Percy Liang]], [[Emily Bender]] va Michael Madden ogohlantirdi: oʻqitishda ishlatilgan maʼlumotlarni bilmay turib, benchmark natijalarini talqin qilish qiyin.<ref name="PeakHype" /><ref name="Conversation" /> ''[[Fast Company]]'' uchun yozgan Mark Sullivan Googlening iPhone bozordagi ustunligiga raqobat qilish imkoniyati borligini taʼkidladi. Uning fikricha, Apple oʻzining [[Siri]] [[virtual assistant|virtual yordamchisi]] bilan Geminiga oʻxshash imkoniyatni yarata olishi dargumon edi.<ref name="AchillesHeel" /> Gemini taqdimotidan keyingi kuni Google aksiyalari 5,3&nbsp;foizga koʻtarildi.<ref name="WallStreetReuters" /><ref name="WallStreetMW" />
 
Google Gemini haqidagi namoyish videosi uchun tanqidga uchradi. Bu video modelning real vaqtdagi muloqot imkoniyatlarini haqiqiy koʻrsatuv sifatida taqdim etilgan edi, aslida esa shunday koʻrinishi uchun tahrir qilingandi. Shunga qaramay, Google boshqa joylarda (video tashqarisida) kichik izohlar berib, uning aslida real vaqtda yozib olinmaganini ogohlantirgan edi.<ref name="jXo97" />
 
Gemini 2.5 Pro Experimental [[LMArena]] reytingida birinchi oʻrinda paydo boʻldi; bu reyting odamlarning afzal koʻrishini oʻlchaydi.<ref name="GoogleBlog25" /><ref name="RDWorld25" /> Model turli sunʼiy intellekt benchmarklarida, jumladan [[Humanity's Last Exam]] sinovida yuqori natijalarga erishdi.<ref name="GoogleBlog25" /><ref name="DeepMindPro25" /><ref name="RDWorld25" /><ref name="TechRadar25" /> Dastlabki sharhlarda uning yaxshilangan mulohaza yuritish qobiliyati va oldingi versiyalarga nisbatan ustunligi qayd etildi.<ref name="TechRadar25" /><ref name="AnalyticsVidhya25" /> Eʼlon qilingan benchmarklar, shuningdek, [[Anthropic]], [[XAI (company)|xAI]] yoki [[OpenAI]] kabi raqobatchilarning oʻsha davrdagi modellari ustun boʻlgan sohalarni ham koʻrsatdi.<ref name="DeepMindPro25" /><ref name="RDWorld25" />
 
== Yana qarang ==
* [[Gato (DeepMind)|Gato]] — DeepMind ishlab chiqqan multimodal neyron tarmoq
* [[Gemini Robotics]]
* [[Reasoning model]]
* [[List of large language models]]
 
== Manbalar ==
{{reflist|refs=
<ref name="IO2023NYT">{{Cite news |last=Grant |first=Nico |date=May 10, 2023 |title=Google Builds on Tech's Latest Craze With Its Own A.I. Products |url=https://www.nytimes.com/2023/05/10/technology/google-ai-products.html |url-access=limited |url-status=live |archive-url=https://web.archive.org/web/20230510180605/https://www.nytimes.com/2023/05/10/technology/google-ai-products.html |archive-date=May 10, 2023 |access-date=August 21, 2023 |newspaper=[[The New York Times]] |issn=0362-4331}}</ref>
<ref name="IO2023ZD">{{Cite web |last=Ortiz |first=Sabrina |date=May 10, 2023 |title=Every major AI feature announced at Google I/O 2023 |url=https://www.zdnet.com/article/every-major-ai-feature-announced-at-google-io-2023/ |url-status=live |archive-url=https://web.archive.org/web/20230510224825/https://www.zdnet.com/article/every-major-ai-feature-announced-at-google-io-2023/ |archive-date=May 10, 2023 |access-date=August 21, 2023 |website=[[ZDNet]]}}</ref>
<ref name="AnnounceGuardian">{{Cite news |last=Milmo |first=Dan |date=December 6, 2023 |title=Google says new AI model Gemini outperforms ChatGPT in most tests |url=https://www.theguardian.com/technology/2023/dec/06/google-new-ai-model-gemini-bard-upgrade |url-status=live |archive-url=https://web.archive.org/web/20231206162533/https://www.theguardian.com/technology/2023/dec/06/google-new-ai-model-gemini-bard-upgrade |archive-date=December 6, 2023 |access-date=December 6, 2023 |newspaper=[[The Guardian]] |issn=0261-3077}}</ref>
<ref name="Google25">{{Cite magazine |last=Levy |first=Steven |author-link=Steven Levy |date=September 11, 2023 |title=Sundar Pichai on Google;s AI, Microsoft's AI, OpenAI, and ... Did We Mention AI? |url=https://www.wired.com/story/sundar-pichai-google-ai-microsoft-openai/ |url-access=limited |url-status=live |archive-url=https://web.archive.org/web/20230911124432/https://www.wired.com/story/sundar-pichai-google-ai-microsoft-openai/ |archive-date=September 11, 2023 |access-date=September 12, 2023 |magazine=[[Wired (magazine)|Wired]]}}</ref>
<ref name="HassabisInterview">{{Cite magazine |last=Knight |first=Will |date=June 26, 2023 |title=Google DeepMind's CEO Says Its Next Algorithm Will Eclipse ChatGPT |url=https://www.wired.com/story/google-deepmind-demis-hassabis-chatgpt/ |url-access=limited |url-status=live |archive-url=https://web.archive.org/web/20230626121231/https://www.wired.com/story/google-deepmind-demis-hassabis-chatgpt/ |archive-date=June 26, 2023 |access-date=August 21, 2023 |magazine=[[Wired (magazine)|Wired]]}}</ref>
<ref name="TheInformation">{{Cite web |last=Victor |first=Jon |date=August 15, 2023 |title=How Google is Planning to Beat OpenAI |url=https://www.theinformation.com/articles/the-forced-marriage-at-the-heart-of-googles-ai-race |url-access=subscription |url-status=live |archive-url=https://web.archive.org/web/20230815134508/https://www.theinformation.com/articles/the-forced-marriage-at-the-heart-of-googles-ai-race |archive-date=August 15, 2023 |access-date=August 21, 2023 |website=[[The Information (website)|The Information]]}}</ref>
<ref name="LarrySergey">{{Cite news |last=Grant |first=Nico |date=January 20, 2023 |title=Google Calls In Help From Larry Page and Sergey Brin for A.I. Fight |url=https://www.nytimes.com/2023/01/20/technology/google-chatgpt-artificial-intelligence.html |url-access=limited |url-status=live |archive-url=https://web.archive.org/web/20230120081118/https://www.nytimes.com/2023/01/20/technology/google-chatgpt-artificial-intelligence.html |archive-date=January 20, 2023 |access-date=February 6, 2023 |newspaper=[[The New York Times]] |issn=0362-4331}}</ref>
<ref name="BrinReturn">{{Cite news |last1=Kruppa |first1=Miles |last2=Seetharaman |first2=Deepa |date=July 21, 2023 |title=Sergey Brin Is Back in the Trenches at Google |url=https://www.wsj.com/articles/sergey-brin-google-ai-gemini-1b5aa41e |url-access=subscription |url-status=live |archive-url=https://archive.today/20230721010618/https://www.wsj.com/amp/articles/sergey-brin-google-ai-gemini-1b5aa41e |archive-date=July 21, 2023 |access-date=September 7, 2023 |newspaper=[[The Wall Street Journal]] |issn=0099-9660}}</ref>
<ref name="CoreContributor">{{Cite web |last=Carter |first=Tom |date=December 7, 2023 |title=Google confirms that its cofounder Sergey Brin played a key role in creating its ChatGPT rival |url=https://www.businessinsider.com/google-sergey-brin-back-from-retirement-to-build-gemini-2023-12 |url-access=limited |url-status=live |archive-url=https://web.archive.org/web/20231207213015/https://www.businessinsider.com/google-sergey-brin-back-from-retirement-to-build-gemini-2023-12 |archive-date=December 7, 2023 |access-date=December 31, 2023 |website=[[Business Insider]]}}</ref>
<ref name="Hustle">{{Cite web |last=Victor |first=Jon |date=September 18, 2023 |title=OpenAI Hustles to Beat Google to Launch 'Multimodal' LLM |url=https://www.theinformation.com/articles/openai-hustles-to-beat-google-to-launch-multimodal-llm |url-access=subscription |url-status=live |archive-url=https://web.archive.org/web/20230918174849/https://www.theinformation.com/articles/openai-hustles-to-beat-google-to-launch-multimodal-llm |archive-date=September 18, 2023 |access-date=October 15, 2023 |website=[[The Information (website)|The Information]]}}</ref>
<ref name="NearLaunch">{{Cite news |author= |date=September 14, 2023 |title=Google nears release of AI software Gemini, The Information reports |url=https://www.reuters.com/technology/google-nears-release-ai-software-gemini-information-2023-09-15/ |url-access=limited |url-status=live |archive-url=https://web.archive.org/web/20230915113109/https://www.reuters.com/technology/google-nears-release-ai-software-gemini-information-2023-09-15/ |archive-date=September 15, 2023 |access-date=October 2, 2023 |publisher=[[Reuters]]}}</ref>
<ref name="WhatWeKnow">{{Cite web |last=Nolan |first=Beatrice |date=September 23, 2023 |title=Google is quietly handing out early demos of its GPT-4 rival called Gemini. Here's what we know so far about the upcoming AI model. |url=https://www.businessinsider.com/google-gemini-explainer-ai-model-2023-9 |url-access=limited |url-status=live |archive-url=https://web.archive.org/web/20230923121028/https://www.businessinsider.com/google-gemini-explainer-ai-model-2023-9 |archive-date=September 23, 2023 |access-date=October 16, 2023 |website=[[Business Insider]]}}</ref>
<ref name="AnnounceWSJ">{{Cite news |last=Kruppa |first=Miles |date=December 6, 2023 |title=Google Announces AI System Gemini After Turmoil at Rival OpenAI |url=https://www.wsj.com/tech/ai/google-announces-ai-system-gemini-after-turmoil-at-rival-openai-10835335 |url-access=subscription |url-status=live |archive-url=https://archive.today/20231206152820/https://www.wsj.com/tech/ai/google-announces-ai-system-gemini-after-turmoil-at-rival-openai-10835335 |archive-date=December 6, 2023 |access-date=December 6, 2023 |newspaper=[[The Wall Street Journal]] |issn=0099-9660}}</ref>
<ref name="AnnounceAP">{{Cite news |last1=Liedtike |first1=Michael |last2=O'Brien |first2=Matt |date=December 6, 2023 |title=Google launches Gemini, upping the stakes in the global AI race |url=https://apnews.com/article/google-gemini-artificial-intelligence-launch-95d05d02051e75e20b574614ae720b8b |url-status=live |archive-url=https://web.archive.org/web/20231206181414/https://apnews.com/article/google-gemini-artificial-intelligence-launch-95d05d02051e75e20b574614ae720b8b |archive-date=December 6, 2023 |access-date=December 6, 2023 |publisher=[[Associated Press]]}}</ref>
<ref name="AnnounceArs">{{Cite web |last=Edwards |first=Benj |date=December 6, 2023 |title=Google launches Gemini—a powerful AI model it says can surpass GPT-4 |url=https://arstechnica.com/information-technology/2023/12/google-launches-gemini-a-powerful-ai-model-it-says-can-surpass-gpt-4/ |url-status=live |archive-url=https://web.archive.org/web/20231206182034/https://arstechnica.com/information-technology/2023/12/google-launches-gemini-a-powerful-ai-model-it-says-can-surpass-gpt-4/ |archive-date=December 6, 2023 |access-date=December 6, 2023 |website=[[Ars Technica]]}}</ref>
<ref name="AnnounceVerge">{{Cite web |last=Pierce |first=David |date=December 6, 2023 |title=Google launches Gemini, the AI model it hopes will take down GPT-4 |url=https://www.theverge.com/2023/12/6/23990466/google-gemini-llm-ai-model |url-status=live |archive-url=https://web.archive.org/web/20231206174404/https://www.theverge.com/2023/12/6/23990466/google-gemini-llm-ai-model |archive-date=December 6, 2023 |access-date=December 6, 2023 |website=[[The Verge]]}}</ref>
<ref name="AnnounceCNN">{{Cite web |last1=Fung |first1=Brian |last2=Thorbecke |first2=Catherine |date=December 6, 2023 |title=Google launches Gemini, its most-advanced AI model yet, as it races to compete with ChatGPT |url=https://www.cnn.com/2023/12/06/tech/google-launches-gemini-compete-with-chatgpt/index.html |url-status=live |archive-url=https://web.archive.org/web/20231206183632/https://www.cnn.com/2023/12/06/tech/google-launches-gemini-compete-with-chatgpt/index.html |archive-date=December 6, 2023 |access-date=December 6, 2023 |publisher=[[CNN Business]]}}</ref>
<ref name="AnnounceCNBC">{{Cite web |last=Elias |first=Jennifer |date=December 6, 2023 |title=Google launches its largest and 'most capable' AI model, Gemini |url=https://www.cnbc.com/2023/12/06/google-launches-its-largest-and-most-capable-ai-model-gemini.html |url-status=live |archive-url=https://web.archive.org/web/20231206160659/https://www.cnbc.com/2023/12/06/google-launches-its-largest-and-most-capable-ai-model-gemini.html |archive-date=December 6, 2023 |access-date=December 6, 2023 |publisher=[[CNBC]]}}</ref>
<ref name="AnnounceCBS">{{Cite web |date=December 6, 2023 |title=Google launches Gemini, upping the stakes in the global AI race |url=https://www.cbsnews.com/sanfrancisco/news/google-ups-the-stakes-in-ai-race-with-gemini-a-technology-trained-to-behave-more-like-humans/ |url-status=live |archive-url=https://web.archive.org/web/20231207001332/https://www.cbsnews.com/sanfrancisco/news/google-ups-the-stakes-in-ai-race-with-gemini-a-technology-trained-to-behave-more-like-humans/ |archive-date=December 7, 2023 |access-date=December 7, 2023 |publisher=[[CBS News]]}}</ref>
<ref name="AnnounceTime">{{Cite web |last=Henshall |first=Will |date=December 6, 2023 |title=Google DeepMind Unveils Its Most Powerful AI Offering Yet |url=https://time.com/6343450/gemini-google-deepmind-ai/ |url-access=limited |url-status=live |archive-url=https://web.archive.org/web/20231206232509/https://time.com/6343450/gemini-google-deepmind-ai/ |archive-date=December 6, 2023 |access-date=December 6, 2023 |magazine=[[Time (magazine)|Time]]}}</ref>
<ref name="AnnounceNYT">{{Cite news |last1=Metz |first1=Cade |last2=Grant |first2=Nico |date=December 6, 2023 |title=Google Updates Bard Chatbot With 'Gemini' A.I. as It Chases ChatGPT |url=https://www.nytimes.com/2023/12/06/technology/google-ai-bard-chatbot-gemini.html |url-status=live |archive-url=https://web.archive.org/web/20231206153133/https://www.nytimes.com/2023/12/06/technology/google-ai-bard-chatbot-gemini.html |archive-date=December 6, 2023 |access-date=December 6, 2023 |newspaper=[[The New York Times]] |issn=0362-4331}}</ref>
<ref name="AnnounceWired">{{Cite magazine |last=Knight |first=Will |date=December 6, 2023 |title=Google Just Launched Gemini, Its Long-Awaited Answer to ChatGPT |url=https://www.wired.com/story/google-gemini-ai-model-chatgpt/ |url-access=limited |url-status=live |archive-url=https://web.archive.org/web/20231206151324/https://www.wired.com/story/google-gemini-ai-model-chatgpt/ |archive-date=December 6, 2023 |access-date=December 6, 2023 |magazine=[[Wired (magazine)|Wired]]}}</ref>
<ref name="AnnounceBloomberg">{{Cite news |last1=Alba |first1=Davey |author-link1=Davey Alba |last2=Ghaffary |first2=Shirin |date=December 6, 2023 |title=Google Opens Access to Gemini, Racing to Catch Up to OpenAI |url=https://www.bloomberg.com/news/articles/2023-12-06/google-opens-access-to-gemini-racing-to-catch-up-to-openai |url-access=limited |url-status=live |archive-url=https://web.archive.org/web/20231206161403/https://www.bloomberg.com/news/articles/2023-12-06/google-opens-access-to-gemini-racing-to-catch-up-to-openai |archive-date=December 6, 2023 |access-date=December 7, 2023 |publisher=[[Bloomberg News]]}}</ref>
<ref name="NewBreed">{{Cite magazine |last=Knight |first=Will |date=December 6, 2023 |title=Google DeepMind's Demis Hassabis Says Gemini Is a New Breed of AI |url=https://www.wired.com/story/google-deepmind-demis-hassabis-gemini-ai/ |url-access=limited |url-status=live |archive-url=https://web.archive.org/web/20231206153212/https://www.wired.com/story/google-deepmind-demis-hassabis-gemini-ai/ |archive-date=December 6, 2023 |access-date=December 7, 2023 |magazine=[[Wired (magazine)|Wired]]}}</ref>
<ref name="S24Bloomberg">{{Cite news |last1=Gurman |first1=Mark |last2=Love |first2=Julia |last3=Alba |first3=Davey |date=January 17, 2024 |title=Samsung Bets on Google-Powered AI Features in Smartphone Revamp |url=https://www.bloomberg.com/news/articles/2024-01-17/samsung-unpacked-event-galaxy-s24-s24-s24-ultra-phones-include-google-ai |url-access=limited |url-status=live |archive-url=https://web.archive.org/web/20240117182045/https://www.bloomberg.com/news/articles/2024-01-17/samsung-unpacked-event-galaxy-s24-s24-s24-ultra-phones-include-google-ai |archive-date=January 17, 2024 |access-date=February 4, 2024 |publisher=[[Bloomberg News]]}}</ref>
<ref name="S24Wired">{{Cite magazine |last=Chokkattu |first=Julian |date=January 17, 2024 |title=Samsung's Galaxy S24 Phones Call on Google's AI to Spruce Up Their Smarts |url=https://www.wired.com/story/samsung-galaxy-unpacked-2024/ |url-access=limited |url-status=live |archive-url=https://web.archive.org/web/20240117181252/https://www.wired.com/story/samsung-galaxy-unpacked-2024/ |archive-date=January 17, 2024 |access-date=February 4, 2024 |magazine=[[Wired (magazine)|Wired]]}}</ref>
<ref name="GeminiRebrandNYT">{{Cite news |last=Metz |first=Cade |date=February 8, 2024 |title=Google Releases Gemini, an A.I.-Driven Chatbot and Voice Assistant |url=https://www.nytimes.com/2024/02/08/technology/google-gemini-ai-app.html |url-access=limited |url-status=live |archive-url=https://web.archive.org/web/20240208132803/https://www.nytimes.com/2024/02/08/technology/google-gemini-ai-app.html |archive-date=February 8, 2024 |access-date=February 8, 2024 |newspaper=[[The New York Times]] |issn=0362-4331}}</ref>
<ref name="GeminiRebrandReuters">{{Cite news |last=Dastin |first=Jeffrey |date=February 8, 2024 |title=Google rebrands Bard chatbot as Gemini, rolls out paid subscription |url=https://www.reuters.com/technology/google-rebrands-bard-chatbot-gemini-rolls-out-paid-subscription-2024-02-08/ |url-access=limited |url-status=live |archive-url=https://web.archive.org/web/20240208162345/https://www.reuters.com/technology/google-rebrands-bard-chatbot-gemini-rolls-out-paid-subscription-2024-02-08/ |archive-date=February 8, 2024 |access-date=February 8, 2024 |publisher=[[Reuters]]}}</ref>
<ref name="AIPremium">{{Cite web |last=Li |first=Abner |date=February 8, 2024 |title=Google One AI Premium is $19.99/mo with Gemini Advanced & Gemini for Workspace |url=https://9to5google.com/2024/02/08/google-one-ai-premium/ |url-status=live |archive-url=https://web.archive.org/web/20240208162504/https://9to5google.com/2024/02/08/google-one-ai-premium/ |archive-date=February 8, 2024 |access-date=February 8, 2024 |website=[[9to5Google]]}}</ref>
<ref name="Global">{{Cite web |last=Mehta |first=Ivan |date=February 1, 2024 |title=Google's Bard chatbot gets the Gemini Pro update globally |url=https://techcrunch.com/2024/02/01/googles-bard-chatbot-gets-the-gemini-pro-update-globally/ |url-status=live |archive-url=https://web.archive.org/web/20240201150753/https://techcrunch.com/2024/02/01/googles-bard-chatbot-gets-the-gemini-pro-update-globally/ |archive-date=February 1, 2024 |access-date=February 4, 2024 |website=[[TechCrunch]]}}</ref>
<ref name="1.5Wired">{{Cite magazine |last=Knight |first=Will |date=February 15, 2024 |title=Google's Flagship AI Model Gets a Mighty Fast Upgrade |url=https://www.wired.com/story/google-deepmind-gemini-pro-ai-upgrade/ |url-access=limited |url-status=live |archive-url=https://web.archive.org/web/20240215152939/https://www.wired.com/story/google-deepmind-gemini-pro-ai-upgrade/ |archive-date=February 15, 2024 |access-date=February 21, 2024 |magazine=[[Wired (magazine)|Wired]]}}</ref>
<ref name="1.5Forbes">{{Cite magazine |last=Nieva |first=Richard |date=February 15, 2024 |title=Google Unveils Gemini 1.5, But Only Developers And Enterprise Clients Have Access For Now |url=https://www.forbes.com/sites/richardnieva/2024/02/14/google-deepmind-gemini/ |url-access=subscription |url-status=live |archive-url=https://web.archive.org/web/20240215182848/https://www.forbes.com/sites/richardnieva/2024/02/14/google-deepmind-gemini/ |archive-date=February 15, 2024 |access-date=February 21, 2024 |magazine=[[Forbes]]}}</ref>
<ref name="1.5FC">{{Cite magazine |last=McCracken |first=Harry |date=February 15, 2024 |title=Google's new Gemini 1.5 AI can dive deep into oceans of video and audio |url=https://www.fastcompany.com/91029527/google-gemini-1-5-ai-llm |url-access=limited |url-status=live |archive-url=https://web.archive.org/web/20240217180943/https://www.fastcompany.com/91029527/google-gemini-1-5-ai-llm |archive-date=February 17, 2024 |access-date=February 21, 2024 |magazine=[[Fast Company]]}}</ref>
<ref name="1.5BI">{{Cite web |last=Stokes |first=Samantha |date=February 15, 2024 |title=Here's everything you need to know about Gemini 1.5, Google's newly updated AI model that hopes to challenge OpenAI |url=https://www.businessinsider.com/google-gemini-1-5-retools-ai-into-one-advanced-model-2024-2 |url-access=limited |url-status=live |archive-url=https://web.archive.org/web/20240219195159/https://www.businessinsider.com/google-gemini-1-5-retools-ai-into-one-advanced-model-2024-2 |archive-date=February 19, 2024 |access-date=February 21, 2024 |website=[[Business Insider]]}}</ref>
<ref name="GemmaFC">{{Cite magazine |last=Khan |first=Jeremy |date=February 21, 2024 |title=Google unveils new family of open-source AI models called Gemma to take on Meta and others—deciding open-source AI ain't so bad after all |url=https://fortune.com/2024/02/21/google-new-family-open-source-ai-models-gemma/ |url-access=limited |url-status=live |archive-url=https://web.archive.org/web/20240221135344/https://fortune.com/2024/02/21/google-new-family-open-source-ai-models-gemma/ |archive-date=February 21, 2024 |access-date=February 21, 2024 |magazine=[[Fast Company]]}}</ref>
<ref name="GemmaBloomberg">{{Cite news |last=Alba |first=Davey |date=February 21, 2024 |title=Google Delves Deeper Into Open Source with Launch of Gemma AI Model |url=https://www.bloomberg.com/news/articles/2024-02-21/google-releases-gemma-ai-model-for-open-source-developers |url-access=limited |url-status=live |archive-url=https://web.archive.org/web/20240221131813/https://www.bloomberg.com/news/articles/2024-02-21/google-releases-gemma-ai-model-for-open-source-developers |archive-date=February 21, 2024 |access-date=February 21, 2024 |publisher=[[Bloomberg News]]}}</ref>
<ref name="GemmaNYT">{{Cite news |last1=Metz |first1=Cade |last2=Grant |first2=Nico |date=February 21, 2024 |title=Google Is Giving Away Some of the A.I. That Powers Chatbots |url=https://www.nytimes.com/2024/02/21/technology/google-open-source-ai.html |url-access=limited |url-status=live |archive-url=https://web.archive.org/web/20240221140434/https://www.nytimes.com/2024/02/21/technology/google-open-source-ai.html |archive-date=February 21, 2024 |access-date=February 21, 2024 |newspaper=[[The New York Times]] |issn=0362-4331}}</ref>
<ref name="Flash">{{Cite web |last=Elias |first=Jennifer |date=August 12, 2024 |title=Google rolls out its most powerful AI models as competition from OpenAI heats up |url=https://www.cnbc.com/2024/05/14/google-announces-lightweight-ai-model-gemini-flash-1point5-at-google-i/o.html |url-status=live |archive-url=https://web.archive.org/web/20240514182111/https://www.cnbc.com/2024/05/14/google-announces-lightweight-ai-model-gemini-flash-1point5-at-google-i/o.html |archive-date=May 14, 2024 |access-date=August 13, 2024 |publisher=[[CNBC]]}}</ref>
<ref name="GoogleBlog25">{{cite web |url=https://blog.google/technology/google-deepmind/gemini-model-thinking-updates-march-2025/ |title=Gemini 2.5: Our most intelligent AI model |website=The Keyword (Google Blog) |date=March 25, 2025 |access-date=March 26, 2025 |last=Kavukcuoglu |first=Koray}}</ref>
<ref name="TechRadar25">{{cite web |url=https://www.techradar.com/computing/artificial-intelligence/gemini-2-5-is-now-available-for-advanced-users-and-it-seriously-improves-googles-ai-reasoning |title=Gemini 2.5 is now available for Advanced users and it seriously improves Google's AI reasoning |website=TechRadar |date=March 26, 2025 |access-date=March 26, 2025}}</ref>
<ref name="RDWorld25">{{cite web |url=https://www.rdworldonline.com/googles-gemini-2-5-pro-model-tops-lmarena-by-40-points-outperforms-competitors-in-scientific-reasoning/ |title=Google's Gemini 2.5 Pro model tops LMArena by close to 40 points |website=R&D World |date=March 25, 2025 |access-date=March 26, 2025}}</ref>
<ref name="AnalyticsVidhya25">{{cite web |url=https://www.analyticsvidhya.com/blog/2025/03/gemini-2-5-pro-experimental/ |title=Gemini 2.5 Pro Exp: How to Access, Features, Applications & More |website=Analytics Vidhya |date=March 26, 2025 |access-date=March 26, 2025}}</ref>
<ref name="WhitePaper">{{Cite tech report |date=December 6, 2023 |title=Gemini: A Family of Highly Capable Multimodal Models |url=https://storage.googleapis.com/deepmind-media/gemini/gemini_1_report.pdf |url-status=live |archive-url=https://web.archive.org/web/20231206151650/https://storage.googleapis.com/deepmind-media/gemini/gemini_1_report.pdf |archive-date=December 6, 2023 |access-date=December 7, 2023 |publisher=[[Google DeepMind]]}}</ref>
<ref name="WhitePaper1.5">{{Cite tech report |date=February 15, 2024 |title=Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context |url=https://storage.googleapis.com/deepmind-media/gemini/gemini_v1_5_report.pdf |url-status=live |archive-url=https://web.archive.org/web/20240226230625/https://storage.googleapis.com/deepmind-media/gemini/gemini_v1_5_report.pdf |archive-date=Feb 26, 2024 |access-date=May 17, 2024 |publisher=[[Google DeepMind]]}}</ref>
<ref name="PeakHype">{{Cite magazine |last1=Heikkilä |first1=Melissa |last2=Heaven |first2=Will Douglas |date=December 6, 2023 |title=Google DeepMind's new Gemini model looks amazing—but could signal peak AI hype |url=https://www.technologyreview.com/2023/12/06/1084471/google-deepminds-new-gemini-model-looks-amazing-but-could-signal-peak-ai-hype/ |url-access=limited |url-status=live |archive-url=https://web.archive.org/web/20231206175747/https://www.technologyreview.com/2023/12/06/1084471/google-deepminds-new-gemini-model-looks-amazing-but-could-signal-peak-ai-hype/ |archive-date=December 6, 2023 |access-date=December 6, 2023 |magazine=[[MIT Technology Review]]}}</ref>
<ref name="Hsiao">{{Cite web |last=Langley |first=Hugh |date=October 12, 2023 |title=Google VP teases Gemini's multimodal future: 'I've seen some pretty amazing things.' |url=https://www.businessinsider.com/google-sissie-hsiao-teases-gemini-ai-model-pretty-amazing-things-2023-10 |url-access=limited |url-status=live |archive-url=https://web.archive.org/web/20231012134647/https://www.businessinsider.com/google-sissie-hsiao-teases-gemini-ai-model-pretty-amazing-things-2023-10 |archive-date=October 12, 2023 |access-date=October 15, 2023 |website=[[Business Insider]]}}</ref>
<ref name="Conversation">{{Cite web |last=Madden |first=Michael G. |date=December 15, 2023 |title=Google's Gemini: is the new AI model really better than ChatGPT? |url=https://theconversation.com/googles-gemini-is-the-new-ai-model-really-better-than-chatgpt-219526 |url-status=live |archive-url=https://web.archive.org/web/20231215140239/https://theconversation.com/googles-gemini-is-the-new-ai-model-really-better-than-chatgpt-219526 |archive-date=December 15, 2023 |access-date=February 4, 2024 |website=[[The Conversation (website)|The Conversation]]}}</ref>
<ref name="AchillesHeel">{{Cite magazine |last=Sullivan |first=Mark |date=December 6, 2023 |title=Gemini-powered Google phones may make Siri even more of an Achilles' heel for the iPhone |url=https://www.fastcompany.com/90994058/gemini-ai-android-google-apple |url-access=limited |url-status=live |archive-url=https://web.archive.org/web/20231207063441/https://www.fastcompany.com/90994058/gemini-ai-android-google-apple |archive-date=December 7, 2023 |access-date=December 7, 2023 |magazine=[[Fast Company]]}}</ref>
<ref name="WallStreetReuters">{{Cite news |last=Soni |first=Aditya |date=December 7, 2023 |title=Alphabet soars as Wall Street cheers arrival of AI model Gemini |url=https://www.reuters.com/technology/alphabet-soars-wall-street-cheers-arrival-ai-model-gemini-2023-12-07/ |url-access=limited |url-status=live |archive-url=https://web.archive.org/web/20231207203954/https://www.reuters.com/technology/alphabet-soars-wall-street-cheers-arrival-ai-model-gemini-2023-12-07/ |archive-date=December 7, 2023 |access-date=February 4, 2024 |publisher=[[Reuters]]}}</ref>
<ref name="WallStreetMW">{{Cite web |last=Swartz |first=Jon |date=December 7, 2023 |title=Gemini, Google's long-awaited answer to ChatGPT, is an overnight hit |url=https://www.marketwatch.com/story/gemini-googles-long-awaited-answer-to-chatgpt-is-an-overnight-hit-a523a817 |url-access=limited |url-status=live |archive-url=https://web.archive.org/web/20231207163911/https://www.marketwatch.com/story/gemini-googles-long-awaited-answer-to-chatgpt-is-an-overnight-hit-a523a817 |archive-date=December 7, 2023 |access-date=February 4, 2024 |website=[[MarketWatch]]}}</ref>
<ref name="DeepMindPro25">{{cite web |url=https://deepmind.google/technologies/gemini/pro/ |title=Gemini Pro - Technical Details and Benchmarks |website=Google DeepMind |date=March 25, 2025 |access-date=March 26, 2025}}</ref>
<ref name="G3Flash">{{cite web | title=Gemini 3 Flash: Frontier intelligence built for speed | date=December 17, 2025 | url=https://blog.google/products/gemini/gemini-3-flash/}}</ref>
<ref name="googleblog">{{cite web |title=Gemini 2.5: Our most intelligent models are getting even better |url=https://blog.google/technology/google-deepmind/google-gemini-updates-io-2025/ |date=2025-05-20 |access-date=2025-05-21}}</ref>
<ref name="Doshi-2025">{{cite web |last1=Doshi |first1=Tulsee |title=We're expanding our Gemini 2.5 family of models |url=https://blog.google/products/gemini/gemini-2-5-model-family-expands/ |website=The Keyword |date=June 17, 2025 |publisher=Google |access-date=12 July 2025}}</ref>
<ref name="Gemini_3_Pro_DeepThink_Announcement">{{Cite web |date=2025-11-18 |title=A new era of intelligence with Gemini 3 |url=https://blog.google/products/gemini/gemini-3/ |access-date=2025-11-18 |website=Google |language=en-us |archive-url=https://web.archive.org/web/20251129051752/https://blog.google/products/gemini/gemini-3/ |archive-date=29 November 2025 |url-status=live}}</ref>
<ref name="G3.1Pro">{{Cite web |date=2026-02-19 |title=Gemini 3.1 Pro: A smarter model for your most complex tasks |url=https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-1-pro/ |access-date=2026-02-19 |website=Google |language=en-us}}</ref>
<ref name="G3.1Pro-sec">{{cite web |last=Klinken |first=Erik |url=https://www.techzine.eu/news/applications/138941/google-launches-gemini-3-1-pro-an-llm-for-complex-reasoning/ |title=Google launches Gemini 3.1 Pro, an LLM for complex reasoning |website=Techzine |date=February 20, 2026 |access-date=April 21, 2026 |url-status=live |archive-url=https://web.archive.org/web/20260228024545/https://www.techzine.eu/news/applications/138941/google-launches-gemini-3-1-pro-an-llm-for-complex-reasoning/ |archive-date=February 28, 2026}}</ref>
<ref name="gemini1">{{cite web |last1=Pichai |first1=Sundar |last2=Hassabis |first2=Demis |title=Introducing Gemini: our largest and most capable AI model |url=https://blog.google/technology/ai/google-gemini-ai |website=The Keyword |date=December 6, 2023 |access-date=12 July 2025}}</ref>
<ref name="gemini1.5">{{cite web |last1=Pichai |first1=Sundar |last2=Hassabis |first2=Demiss |date=February 15, 2024 |title=Our next-generation model: Gemini 1.5 |url=https://blog.google/technology/ai/google-gemini-next-generation-model-february-2024/ |access-date=12 July 2025 |website=The Keyword}}</ref>
<ref name="elpais2">{{Cite web |last=País |first=El |date=2025-08-29 |title=Google se vuelve viral con Nano Banana, su modelo más avanzado de edición de imágenes con IA |url=https://elpais.com/tecnologia/2025-08-29/google-se-vuelve-viral-con-nano-banana-su-modelo-mas-avanzado-de-edicion-de-imagenes-con-ia.html |access-date=2025-09-10 |website=El País |language=es}}</ref>
<ref name="gblog2">{{Cite web |date=2025-08-26 |title=Image editing in Gemini just got a major upgrade |url=https://blog.google/products/gemini/updated-image-editing-model/ |access-date=2025-09-10 |website=Google |language=en-us}}</ref>
<ref name="pcgamer2">{{Cite news |last=Kinghorn |first=Jess |date=2025-08-28 |title=Gemini's 'Nano Banana' AI image editor can't crop a picture, but its penchant for deepfakes 'while keeping you, you' makes me want to wear a brown paper bag on my head forever more |url=https://www.pcgamer.com/software/ai/geminis-nano-banana-update-aims-to-keep-people-looking-the-same-in-ai-art-and-the-fear-of-deepfakes-makes-me-want-to-wear-a-brown-paper-bag-on-my-head-forever-more/ |access-date=2025-09-10 |work=PC Gamer |language=en}}</ref>
<ref name="auto32">{{Cite news |date=2025-09-06 |title=Google's Nano Banana available on X: Here's how to use |url=https://economictimes.indiatimes.com/tech/artificial-intelligence/googles-nano-banana-arrives-on-x-heres-how-to-use/articleshow/123736064.cms?from=mdr |access-date=2025-09-10 |work=The Economic Times |issn=0013-0389}}</ref>
<ref name="kf0Qv">[https://blog.google/technology/developers/introducing-gemini-cli-open-source-ai-agent/ Google announces Gemini CLI: your open-source AI agent]. Blog.google. Retrieved 28 August 2025.</ref>
<ref name="c65D6">{{Cite web |title=Updated production-ready Gemini models, reduced 1.5 Pro pricing, increased rate limits, and more- Google Developers Blog |url=https://developers.googleblog.com/en/updated-production-ready-gemini-models-reduced-15-pro-pricing-increased-rate-limits-and-more/ |access-date= |website=developers.googleblog.com |language=en}}</ref>
<ref name="SLeze">{{Cite web |date=2024-12-11 |title=Introducing Gemini 2.0: our new AI model for the agentic era |url=https://blog.google/technology/google-deepmind/google-gemini-ai-update-december-2024/#gemini-2-0 |access-date=2024-12-20 |website=Google |language=en-us}}</ref>
<ref name="S4jSs">{{Cite web |title=The next chapter of the Gemini era for developers- Google Developers Blog |url=https://developers.googleblog.com/en/the-next-chapter-of-the-gemini-era-for-developers/ |access-date=2024-12-20 |website=developers.googleblog.com |language=en}}</ref>
<ref name="UEHpr">{{Cite web |title=Gemini 2.0 Flash (experimental) {{!}} Gemini API |url=https://ai.google.dev/gemini-api/docs/models/gemini-v2 |access-date=2024-12-20 |website=Google AI for Developers |language=en}}</ref>
<ref name="MOz3i">{{cite web |title=Gemini 2.0 is now available to everyone |url=https://blog.google/technology/google-deepmind/gemini-model-updates-february-2025/ |website=Google |language=en-us |date=5 February 2025}}</ref>
<ref name="h3GOR">{{cite web |last1=Parada |first1=Carolina |title=Gemini Robotics brings AI into the physical world |url=https://deepmind.google/discover/blog/gemini-robotics-brings-ai-into-the-physical-world/ |website=Google DeepMind |date=March 12, 2025 |access-date=11 July 2025}}</ref>
<ref name="8KRAo">{{Cite web |last=Abner |first=Li |date=Mar 13, 2025 |title=Gemini in Android Studio can now turn UI mockups into code |url=https://9to5google.com/2025/03/13/gemini-android-studio-images-code/ |access-date=April 12, 2025 |website=[[9to5Google]]}}</ref>
<ref name="k9kC2">{{cite web |title=Deep Think boosts the performance of Google's flagship Gemini AI model |url=https://techcrunch.com/2025/05/20/deep-think-boosts-the-performance-of-googles-flagship-google-gemini-ai-model/ |date=2025-05-20 |access-date=2025-05-21 |website=TechCrunch}}</ref>
<ref name="pzb1P">{{Cite web |last=Warren |first=Tom |date=December 5, 2025 |title=OpenAI's GPT-5.2 'code red' response to Google is coming next week |url=https://www.theverge.com/report/838857/openai-gpt-5-2-release-date-code-red-google-response |access-date=December 9, 2025 |website=The Verge}}</ref>
<ref name="8bOyM">{{Cite web |last=Hart |first=Robert |date=December 2, 2025 |title=OpenAI declares 'code red' as Google catches up in AI race |url=https://www.theverge.com/news/836212/openai-code-red-chatgpt |access-date=December 9, 2025 |website=The Verge}}</ref>
<ref name="twyqf">{{Cite web |last=Edwards |first=Benj |date=December 11, 2025 |title=OpenAI releases GPT-5.2 after "code red" Google threat alert |url=https://arstechnica.com/information-technology/2025/12/openai-releases-gpt-5-2-after-code-red-google-threat-alert/ |access-date=December 15, 2025 |website=Ars Technica}}</ref>
<ref name="0wlt5">{{Cite web |last=Bellan |first=Rebecca |date=December 11, 2025 |title=OpenAI fires back at Google with GPT-5.2 after 'code red' memo |url=https://techcrunch.com/2025/12/11/openai-fires-back-at-google-with-gpt-5-2-after-code-red-memo/ |access-date=December 15, 2025 |website=TechCrunch}}</ref>
<ref name="a7HKh">{{Cite web |date=2025-12-04 |title=Gemini 3 Deep Think is now available in the Gemini app. |url=https://blog.google/products/gemini/gemini-3-deep-think/ |access-date=2025-12-05 |website=Google |language=en-us}}</ref>
<ref name="9QWuW">{{Cite web |date=2025-12-17 |title=Gemini 3 Flash: frontier intelligence built for speed |url=https://blog.google/products/gemini/gemini-3-flash/ |access-date=2025-12-17 |website=Google |language=en-us}}</ref>
<ref name="1wUTT">{{Cite news |last=Duffy |first=Clare |date=2026-01-12 |title=Apple teams up with Google Gemini for AI-powered Siri |url=https://www.cnn.com/2026/01/12/tech/apple-google-gemini-siri |access-date=2026-01-15 |website=CNN |language=en}}</ref>
<ref name="zOJDk">{{Cite web |last=Murti |first=Lola |date=2026-02-26 |title=Google launches Nano Banana 2, updating its viral AI image generator |url=https://www.cnbc.com/2026/02/26/google-launches-nano-banana-2-updating-its-viral-ai-image-generator.html |access-date=2026-02-28 |website=CNBC |language=en}}</ref>
<ref name="v0a1t">{{Cite web |date=2026-03-03 |title=Gemini 3.1 Flash-Lite: Built for intelligence at scale |url=https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-1-flash-lite/ |access-date=2026-03-03 |website=Google |language=en-us}}</ref>
<ref name="vAd8z">{{Cite web |last=Kasanmascheff |first=Markus |date=2026-03-03 |title=Google Launches Gemini 3.1 Flash-Lite for Enterprise Scale |url=https://winbuzzer.com/2026/03/03/google-gemini-31-flash-lite-enterprise-scale-xcxwbn/ |access-date=2026-03-03 |website=WinBuzzer |language=en-US}}</ref>
<ref name="mLYmE">{{Cite web |title=Gemini Release updates |url=https://gemini.google.com/updates |website=Google |access-date=April 9, 2025 |archive-date=April 9, 2025 |archive-url=https://web.archive.org/web/20250409083428/https://gemini.google.com/updates |url-status=live}}</ref>
<ref name="4xunj">{{Cite web |title=Gemini models|url=https://ai.google.dev/gemini-api/docs/models |website=Google |access-date=April 9, 2025 |archive-date=April 9, 2025 |archive-url=https://web.archive.org/web/20250409083428/https://ai.google.dev/gemini-api/docs/models |url-status=live}}</ref>
<ref name="dDYsD">{{cite web |title=Model versions and lifecycle |url=https://cloud.google.com/vertex-ai/generative-ai/docs/learn/model-versions |website=Google Cloud}}</ref>
 
<ref name="ZRrBE">{{cite web |last1=Pichai |first1=Sundar |last2=Hassabis |first2=Demis |last3=Kavukcuoglu |first3=Koray |title=Introducing Gemini 2.0: our new AI model for the agentic era |url=https://blog.google/technology/google-deepmind/google-gemini-ai-update-december-2024 |website=The Keyword |date=December 11, 2024 |access-date=12 July 2025}}</ref>
<ref name="w5kMd">{{cite web |last1=Mallick |first1=Shrestha |last2=Kilpatrick |first2=Logan |title=Gemini 2.0: Flash, Flash-Lite and Pro |url=https://developers.googleblog.com/en/gemini-2-family-expands/ |website=Google for Developers |access-date=12 July 2025}}</ref>
<ref name="OSdTC">https://discuss.ai.google.dev/t/migrate-from-gemini-3-pro-preview-to-gemini-3-1-pro-preview-before-march-9-2026/127062</ref>
<ref name="t5REb">[https://storage.googleapis.com/deepmind-media/Model-Cards/Gemini-3-Pro-Model-Card.pdf <nowiki>[Gemini 3 Pro] External Model Card - November 18, 2025 - v5</nowiki>]</ref>
<ref name="wWwmU">{{cite web | title=Gemini 3 Deep Think is now available in the Gemini app | date=December 4, 2025 | url=https://blog.google/products/gemini/gemini-3-deep-think/}}</ref>
<ref name="S5d8b">{{cite web | title=Advanced version of Gemini with Deep Think officially achieves gold-medal standard at the International Mathematical Olympiad | url=https://deepmind.google/blog/advanced-version-of-gemini-with-deep-think-officially-achieves-gold-medal-standard-at-the-international-mathematical-olympiad/}}</ref>
<ref name="x46Kx">{{Cite web |date=2025-11-20 |title=Introducing Nano Banana Pro |url=https://blog.google/technology/ai/nano-banana-pro/ |access-date=2025-11-20 |website=Google |language=en-us}}</ref>
<ref name="kGaLa">{{Cite web |last=Griffiths |first=Brent D. |title=She named Google's Nano Banana. She has an idea why it went viral. |url=https://www.businessinsider.com/google-nano-banana-name-origin-2026-1 |access-date=2026-01-20 |website=Business Insider |language=en-US}}</ref>
<ref name="vv1gn">{{Cite web |date=2025-11-20 |title=Build with Nano Banana Pro, our Gemini 3 Pro Image model |url=https://blog.google/technology/developers/gemini-3-pro-image-developers/ |access-date=2025-11-20 |website=Google |language=en-us}}</ref>
<ref name="DtCVU">{{Cite web |date=2025-11-20 |title=7 tips to get the most out of Nano Banana Pro |url=https://blog.google/products/gemini/prompting-tips-nano-banana-pro/ |access-date=2025-11-20 |website=Google |language=en-us}}</ref>
<ref name="dQrC4">{{Cite web |last=Li |first=Abner |date=2025-09-04 |title=Nano Banana responsible for 10+ million first-time Gemini app users |url=https://9to5google.com/2025/09/04/gemini-app-nano-banana/ |access-date=2025-09-10 |website=9to5Google |language=en-US}}</ref>
<ref name="dd94J">{{Cite web |date=2025-09-05 |title=Google celebrates major achievements by its Gemini app 'Nano Banana' editor |url=https://www.androidcentral.com/apps-software/ai/google-says-nano-banana-drove-in-over-10-million-new-users-to-gemini-app |access-date=2025-09-10 |website=Android Central |language=en}}</ref>
<ref name="HHNth">{{Cite news |date=2025-09-10 |title=Nano Banana' trend: What is the viral 3D figurine craze, and how to create one for free |url=https://timesofindia.indiatimes.com/etimes/trending/nano-banana-trend-what-is-the-viral-3d-figurine-craze-and-how-to-create-one-for-free/articleshow/123785545.cms |access-date=2025-09-10 |work=The Times of India |issn=0971-8257}}</ref>
<ref name="v1ey9">{{Cite web |date=2025-09-01 |title=Gemini's Nano Banana makes photo editing fun again – and yes, it's better than ChatGPT |url=https://www.techradar.com/ai-platforms-assistants/gemini/i-spent-the-weekend-comparing-geminis-new-nano-banana-image-tool-to-chatgpt-and-theres-one-clear-winner |access-date=2025-09-10 |website=TechRadar |language=en}}</ref>
<ref name="JFRoK">{{Cite web |last=Caswell |first=Amanda |date=2025-09-09 |title=Nano Banana just broke the internet with these viral trends — I tried these 5 prompts and I'm blown away |url=https://www.tomsguide.com/ai/ai-image-video/nano-banana-just-broke-the-internet-with-these-viral-trends-here-are-5-ai-photo-prompts-to-try-now |access-date=2025-09-10 |website=Tom's Guide |language=en}}</ref>
<ref name="lLFT0">{{Cite web |last=Nelson |first=Michael |date=2025-09-11 |title=Nano-Banana and Architectural Visualization for Adaptive Reuse Projects |url=https://amicoarchitectural.com/architectural-visualization-for-adaptive-reuse/ |access-date=2025-09-11 |website=AMICO Architectural Metals |language=en-US}}</ref>
<ref name="q3e3Z">{{cite web |title=Our next-generation model: Gemini 1.5 |url=https://blog.google/technology/ai/google-gemini-next-generation-model-february-2024/#context-window |website=Google |language=en-us |date=15 February 2024 |quote=In our research, we've also successfully tested up to 10 million tokens.}}</ref>
<ref name="jXo97">{{Cite web |last1=Kovach |first1=Steve |last2=Elias |first2=Jennifer |date=2023-12-08 |title=Google Faces Controversy over Edited Gemini AI Demo Video |url=https://www.cnbc.com/2023/12/08/google-faces-controversy-over-edited-gemini-ai-demo-video.html |url-status=live |archive-date=December 9, 2023 |archive-url=https://web.archive.org/web/20231209010937/https://www.cnbc.com/2023/12/08/google-faces-controversy-over-edited-gemini-ai-demo-video.html |publisher=[[CNBC]] |access-date=2023-12-09 |language=en}}</ref>
<ref name="WmcRt">https://blog.google/innovation-and-ai/technology/developers-tools/gemma-4/</ref>
}}
 
== Qoʻshimcha adabiyotlar ==
{{Refbegin}}
* {{Cite magazine |last=Honan |first=Mat |date=December 6, 2023 |title=Google CEO Sundar Pichai on Gemini and the Coming Age of AI |url=https://www.technologyreview.com/2023/12/06/1084539/google-ceo-sundar-pichai-on-gemini-and-the-coming-age-of-ai/ |url-access=subscription |url-status=live |archive-url=https://web.archive.org/web/20231206233355/https://www.technologyreview.com/2023/12/06/1084539/google-ceo-sundar-pichai-on-gemini-and-the-coming-age-of-ai/ |archive-date=December 6, 2023 |magazine=[[MIT Technology Review]] |access-date=December 6, 2023}}
* {{Cite web |last1=Ji |first1=Ju-yeong |last2=Kumar |first2=Ravin |access-date=2024-08-15 |title=Gemma Explained: An Overview of Gemma Model Family Architectures |url=https://developers.googleblog.com/en/gemma-explained-overview-gemma-model-family-architectures/ |website=Google Developers Blog |language=en}}
{{Refend}}
 
== Tashqi havolalar ==
* {{Official website}}
* [https://blog.google/technology/ai/google-gemini-ai/ Matbuot xabari] — ''[[The Keyword]]'' orqali
* Oq kitob: [https://storage.googleapis.com/deepmind-media/gemini/gemini_1_report.pdf 1.0] va [https://storage.googleapis.com/deepmind-media/gemini/gemini_v1_5_report.pdf 1.5]
* [https://storage.googleapis.com/deepmind-media/Model-Cards/Gemini-3-Pro-Model-Card.pdf Gemini 3 Pro uchun oq kitob]
 
{{Google AI}}
{{Google LLC}}
{{Generative AI}}
{{Artificial intelligence navbox}}
 
[[Turkum:yilgi dasturiy taʼminot]]
[[Turkum:Chatbotlar]]
[[Turkum:Google DeepMind]]
[[Turkum:Google dasturlari]]
[[Turkum:Katta til modellari]]
[[Turkum:Multimodal oʻzaro taʼsir]]
[[Turkum:yilda sunʼiy intellekt]]
[[Turkum:Generativ oldindan oʻqitilgan transformerlar]]

$2, $3 dagi (joriy) koʻrinishi

Gemini
Yaratgan dasturchisi (firmasi) Google AI
Google DeepMind
Dastlabki foydalanishga
chiqarilishi		 6-dekabr, 2023-yil (2 yil avval) (2023-12-06) (beta version)
8-fevral, 2024-yil (2 yil avval) (2024-02-08) (official rollout)
Dasturning amaliyotga tatbiqi
3.1 Pro, 3 Deep Think, 3.8 Flash, 3.5 Flash-Lite [1][2] / 2-sentyabr, 2026-yil (2 oy avval) (2026-09-02) (3.8 Flash)
Oldingi versiyalari PaLM
Raqiblari English and 70+ other languages

Gemini — Google DeepMind tomonidan ishlab chiqilgan multimodal katta til modellari (LLM) oilasi boʻlib, LaMDA va PaLM 2 modellarining vorisi hisoblanadi. U Gemini Pro, Gemini Deep Think, Gemini Flash va Gemini Flash Lite modellaridan iborat.[1] Model 2023-yil 6-dekabrda eʼlon qilingan. Gemini Gemini suhbatdoshiga (chatbot) asos boʻladi. Uning nomi Egizaklar burji sharafiga qoʻyilgan.

Tarixi

[tahrir | manbasini tahrirlash]

Ishlab chiqilishi

[tahrir | manbasini tahrirlash]

Google kompaniyasi oʻzining shoʻʼba korxonasi Google DeepMind ishlab chiqqan katta til modeli Geminini 2023-yil 10-mayda Google I/O konferensiyasining asosiy maʼruzasida eʼlon qildi. Gemini shu tadbirda taqdim etilgan PaLM 2 modelining yanada qudratli vorisi sifatida koʻrsatildi. Google bosh direktori Sundar Pichai Gemini hali ishlab chiqishning dastlabki bosqichida ekanini aytdi.[3][4] Boshqa katta til modellaridan farqli oʻlaroq, Gemini faqat matn korpusi asosida oʻqitilmagan. U multimodal boʻlishi uchun loyihalangan, yaʼni bir vaqtning oʻzida matn, tasvir, audio, video va dastur kodi kabi turli xil maʼlumot turlarini qayta ishlay oladi.[5] Model Google DeepMind tarkibiga birlashtirilgan Google kompaniyasining ikki boʻlimi — DeepMind va Google Brain hamkorligida yaratilgan.[6] Wired nashriga bergan intervyusida DeepMind bosh direktori Demis Hassabis Gemininig ilgʻor imkoniyatlarini taʼrifladi. Uning fikricha, bu imkoniyatlar algoritmga OpenAI kompaniyasining GPT-4 asosida ishlaydigan ChatGPT xizmatidan ustun kelish imkonini beradi. ChatGPTning ortib borayotgan mashhurligiga Google LaMDA va Bard bilan qattiq raqobat qilib kelayotgan edi. Hassabis DeepMindning AlphaGo dasturi yutuqlarini alohida taʼkidladi. Bu dastur 2016-yilda Go chempioni Li Sedolni magʻlub etib, butun dunyo eʼtiborini tortgandi. Uning soʻzlariga koʻra, Gemini AlphaGo va Google–DeepMindning boshqa katta til modellari qudratini oʻzida birlashtiradi.[7]

2023-yil avgustda The Information nashri Googlening Gemini boʻyicha rejalari haqida maʼruza eʼlon qildi. Unda kompaniya modelni 2023-yil oxirida ishga tushirishni moʻljallagani aytilgandi. Maʼruzaga koʻra, Google koʻpchilik katta til modellariga xos suhbat imkoniyatlarini sunʼiy intellekt yordamida tasvir yaratish bilan birlashtirib, OpenAI va boshqa raqobatchilardan oʻzib ketishga umid qilgan. Bu esa modelga kontekstga mos tasvirlar yaratish va koʻproq qoʻllash holatlariga moslashish imkonini bergan boʻlardi.[8] Bard singari,[9] Google asoschilaridan biri Sergey Brin ham nafaqadan Gemini ustida ishlash uchun qaytarildi. U bilan birga Google Brain va DeepMinddan yuzlab muhandislar jalb qilindi;[8][10] keyinchalik Brin Gemininig «asosiy hissa qoʻshuvchisi» sifatida qayd etildi.[11] Gemini YouTube videolarining matn transkriptlari asosida oʻqitilgani bois, mualliflik huquqi bilan himoyalangan materiallarni saralab tashlash uchun huquqshunoslar jalb etildi.[8]

Gemini yaqinda ishga tushishi haqidagi xabarlardan soʻng, OpenAI GPT-4 ni Geminiga oʻxshash multimodal imkoniyatlar bilan birlashtirish ishlarini tezlashtirdi.[12] The Information sentabr oyida bir necha kompaniyaga katta til modelining «erta versiyasi»dan foydalanishga ruxsat berilgani haqida xabar berdi. Google bu modelni mijozlarga Google Cloudning Vertex AI xizmati orqali taqdim etishni rejalashtirgandi. Nashr yana Google Geminini ham GPT-4, ham Microsoft kompaniyasining GitHub Copilot xizmatiga raqib qilib tayyorlayotganini yozdi.[13][14]

Ishga tushirilishi

[tahrir | manbasini tahrirlash]

2023-yil 6-dekabrda Pichai va Hassabis onlayn matbuot anjumanida «Gemini 1.0» ni eʼlon qildi.[15][16] U uchta modeldan iborat: «juda murakkab vazifalar» uchun moʻljallangan Gemini Ultra; «keng koʻlamli vazifalar» uchun moʻljallangan Gemini Pro; «qurilma ichidagi vazifalar» uchun moʻljallangan Gemini Nano. Ishga tushirilgan paytda Gemini Pro Bard tarkibiga, Gemini Nano esa Pixel 8 Pro smartfoniga integratsiya qilingandi. Gemini Ultra «Bard Advanced»ga asos boʻlishi va 2024-yil boshida dasturchilar uchun ochilishi rejalashtirilgandi. Google Geminini yana Search, Ads, Chrome, Google Workspacedagi Duet AI va AlphaCode 2 mahsulotlariga joriy etmoqchi edi.[17][16] Model faqat ingliz tilida taqdim etildi.[16][18] Kompaniya Geminini oʻzining «eng yirik va eng qudratli sunʼiy intellekt modeli» deb taʼrifladi hamda uni inson xatti-harakatlarini taqlid qiladigan qilib loyihalaganini aytdi.[19][16][20] Shu bilan birga, «keng qamrovli xavfsizlik sinovlari» zarurligi sababli model keyingi yilgacha ommaga taqdim etilmasligi maʼlum qilindi.[15] Gemini Googlening Tensor Processing Unit (TPU) protsessorlarida oʻqitilgan va shular asosida ishlaydi.[15][18] Modelning nomi DeepMind va Google Brain birlashuviga, shuningdek, NASAning Gemini loyihasiga ishoradir.[21]

Gemini Ultra bir qator soha koʻrsatkichlarida GPT-4, Anthropic kompaniyasining Claude 2, Inflection AI kompaniyasining Inflection-2, Meta kompaniyasining LLaMA 2 va xAI kompaniyasining Grok 1 modellaridan ustun kelgani aytildi.[22][15] Gemini Pro esa GPT-3.5dan ustun chiqqani qayd etildi.[5] Bundan tashqari, Gemini Ultra 57 ta fanni qamrab oluvchi Massive Multitask Language Understanding (MMLU) testida inson mutaxassislaridan ustun kelgan birinchi til modeli boʻldi va 90% natija koʻrsatdi.[5][21] Gemini Pro 13-dekabrda Google Cloud mijozlariga AI Studio va Vertex AI orqali taqdim etildi. Gemini Nano esa Android dasturchilari uchun ham ochiladi.[23][24][25] Hassabis yana DeepMind Gemini «dunyo bilan jismonan oʻzaro taʼsirlashish uchun robototexnika bilan qanday birlashtirilishi mumkinligini» oʻrganayotganini maʼlum qildi.[26] AQSh prezidenti Jo Bayden oktabr oyida imzolagan farmonga muvofiq, Google Gemini Ultra sinov natijalarini AQSh federal hukumati bilan boʻlishishini bildirdi. Shu tariqa kompaniya noyabr oyida Bletchli Parkda oʻtkazilgan Sunʼiy intellekt xavfsizligi sammitida belgilangan tamoyillarga rioya qilish yuzasidan Buyuk Britaniya hukumati bilan muzokaralar olib bordi.[5]

2025-yil iyun oyida Google Gemini CLI dasturini taqdim etdi. Bu — Gemini imkoniyatlarini bevosita terminalga olib keluvchi ochiq kodli sunʼiy intellekt agenti. U ilgʻor kod yozish, avtomatlashtirish va muammo yechish vositalarini taklif etadi hamda yakka dasturchilar uchun keng bepul foydalanish chegaralarini beradi.[27]

Yangilanishlar

[tahrir | manbasini tahrirlash]

2025-yil yanvar oyida Google Samsung bilan hamkorlik qilib, Gemini Nano va Gemini Proni Galaxy S24 smartfonlari qatoriga joriy etdi.[28][29] Keyingi oyda Bard va Duet AI Gemini brendi ostida birlashtirildi.[30][31] «Gemini Advanced with Ultra 1.0» esa Google One obuna xizmatining yangi «AI Premium» tarifi orqali chiqarildi.[32] Gemini Pro ham global miqyosda ishga tushirildi.[33]

2024-yil fevral oyida Google Gemini 1.5 modelini taqdim etib, uni 1.0 Ultradan ham qudratli va imkoniyatliroq model deb taʼrifladi.[34][35][36] Oʻzgarishlar orasida yangi arxitektura, mutaxassislar aralashmasi yondashuvi va bir million tokenli kattaroq kontekst oynasi bor.[37] Shu oyda Google Gemma modellarini taqdim etdi — bu Gemini modellarining kichikroq, erkin va ochiq kodli turkumi. Bir qancha nashrlar buni Meta va boshqa kompaniyalarning sunʼiy intellekt modellarini ochiq kodga oʻtkazishiga javob deb baholadi. Ular buni Googlening sunʼiy intellektni yopiq saqlash boʻyicha oldingi amaliyotidan voz kechishi sifatida ham izohladi.[38][39][40] Google 14-mayda 2024-yilgi I/O asosiy maʼruzasida yana bir model — Gemini 1.5 Flashni eʼlon qildi.[41] Shu tadbirda Google Gemini Nanoning ish stoli uchun moslashtirilgan versiyasini «Built-in AI» arxitekturasi orqali bevosita Google Chrome brauzeriga joriy etish rejasini maʼlum qildi. Bu Prompt API («window.ai») kabi tajriba bosqichidagi API vositalari orqali veb-dasturchilarga mahalliy hisoblash imkoniyatlarini ochib beradi.

Gemininig ikkita yangilangan modeli — Gemini-1.5-Pro-002 va Gemini-1.5-Flash-002 2024-yil 24-sentabrda chiqarildi.[42][non-primary source needed]

2024-yil 11-dekabrda Google yangi Gemini 2.0 Flash Experimental modelini eʼlon qildi.[43] Uning imkoniyatlari orasida real vaqtdagi audio va video muloqot uchun Multimodal Live API, ichki tasvir yaratish va boshqariladigan matndan nutq hosil qilish (raqamli belgi bilan), shuningdek, integratsiyalangan Google Search bor.[44] Model yana takomillashgan agentlik imkoniyatlarini, yangi Google Gen AI SDK toʻplamini va GitHub uchun tajriba bosqichidagi «Jules» nomli sunʼiy intellekt kod yozish agentini taqdim etadi.[45][non-primary source needed]

2025-yil 30-yanvarda Google Gemini 2.0 Flashni yangi standart model sifatida chiqardi, Gemini 1.5 Flash esa foydalanish uchun ochiq qoldi. Shundan soʻng 2025-yil 5-fevralda Gemini 2.0 Pro taqdim etildi. Bundan tashqari, Google Gemini 2.0 Flash Thinking Experimental modelini chiqardi. Bu model soʻrovlarga javob berayotganda til modelining fikrlash jarayoni haqida qisqacha maʼlumot hosil qiladi.[46][non-primary source needed]

2025-yil 12-martda Google Gemini Robotics modelini ham eʼlon qildi. Bu Gemini 2.0 modellari oilasiga asoslangan koʻrish-til-harakat modeli.[47][non-primary source needed] Ertasi kuni Google Android Studiodagi Gemini oddiy interfeys maketlarini tushunib, ularni ishlaydigan Jetpack Compose kodiga aylantira olishini maʼlum qildi.[48]

Gemini 2.5 Pro Experimental 2025-yil 25-martda chiqarildi.[49][50][51] Bu model tugʻma multimodallikni saqlagan holda fikrlash zanjiri imkoniyatini taklif etadi.[49][51][52]

Google I/O 2025 tadbirida Google Gemini 2.5 Flash standart model boʻlishini eʼlon qildi.[53] Gemini 2.5 Pro esa eng ilgʻor Gemini modeli sifatida taqdim etildi. U yaxshilangan kod yozish imkoniyatlari va boshqa takomillashuvlarga ega.[54] Ham 2.5 Pro, ham 2.5 Flash tugʻma audio chiqishni qoʻllab-quvvatlaydi.

2025-yil 17-iyunda Google 2.5 Pro va 2.5 Flashning umumiy foydalanishga ochilganini eʼlon qildi. Shu kuni ular Gemini 2.5 Flash-Lite modelini ham taqdim etdi — bu tezlik va narx samaradorligi uchun moslashtirilgan model.[55][non-primary source needed]

2025-yil 18-noyabrda Google 3 Pro va 3 Deep Think modellarining chiqqanini eʼlon qildi.[56] Bu yangi modellar 2.5 Pro va 2.5 Flash oʻrnini egalladi. Mazkur chiqarilish OpenAI kompaniyasini raqib model GPT-5.2 ni tezroq chiqarishga undadi; u 11-dekabrda taqdim etildi.[57][58][59][60]

2025-yil 4-dekabrda Google 3 Deep Think Ultra obunachilariga bosqichma-bosqich yetkazila boshlashini eʼlon qildi.[61][non-primary source needed]

2025-yil 17-dekabrda Google 3 Flash modeli chiqqanini eʼlon qildi; u 2.5 Flashning joriy versiyasi oʻrnini egalladi.[62][non-primary source needed]

2026-yil 12-yanvarda Apple Siri yordamchisining kelgusi versiyasida Gemini sunʼiy intellekt modelidan foydalanish rejasini eʼlon qildi.[63]

2026-yil 11-fevralda Gemini ayollar moda brendi L'Agence bilan hamkorlik qildi: Nyu-York moda haftaligida kompaniyaning virtual kiyib koʻrish texnologiyasidan foydalanildi.[64]

2026-yil 19-fevralda Google Gemini 3.1 Pro modelini chiqardi.[2][65]

2026-yil 26-fevralda Nano Banana 2 taqdim etildi. Bu Gemini 3.1 Flash Image platformasi asosida qurilgan yangilangan versiya.[66]

2026-yil 3-martda Google Gemini 3.1 Flash Lite modelini Google API orqali dasturchilarga taqdim etdi.[67][68]

2026-yil 2-aprelda Google Gemma 4 modelini chiqardi. U ilgʻor mulohaza yuritish va agentlik ish jarayonlari uchun maxsus yaratilgan. [69]

2026-yil 19-mayda Google Gemini 3.5 Flash modelini taqdim etdi — bu 3.5 modellari oilasidagi birinchi ommaviy nashr. [70]

2026-yil 21-iyulda Google Gemini 3.6 Flash va Gemini 3.5 Flash-Lite modellarini chiqardi.[71][72]

Eng yaxshi 20 ta OpenRouter katta til modeli

2026-yil 13-avgustda Google Gemini 3.7 Flash modelini chiqardi.[73]

2026-yil 2-sentabrda Google Gemini 3.8 Flash va 3.8 Flash Cyber modellarini chiqardi.[74]


Model versiyalari

[tahrir | manbasini tahrirlash]
Yana qarang: Gemma (language model)

Quyidagi jadvalda Gemininig asosiy model versiyalari keltirilgan hamda har bir versiyadagi muhim oʻzgarishlar tavsiflangan:[75][76][77]

Version Release date Status[78][55] Description
1.0 Nano 2023-12-06.11.2026 Old version, no longer maintained: Discontinued Designed for on-device tasks and first available in Google's Pixel 8 Pro.[79]
1.0 Pro 2023-12-13.11.2026 Old version, no longer maintained: Discontinued Designed for a diverse range of tasks.[79]
1.0 Ultra 2024-02-08.11.2026 Old version, no longer maintained: Discontinued Google's most powerful offering in the Gemini 1.0 family.[79]
1.5 Pro 2024-02-15.11.2026 Old version, no longer maintained: Discontinued As a successor to the 1.0 series of models, 1.5 Pro offers significantly increased context size (up to 1 million tokens). It is designed to be the most capable model in the Gemini 1.5 family.[80]
1.5 Flash 2024-05-14.11.2026 Old version, no longer maintained: Discontinued This version got renamed from 'Nano' to 'Flash'. It is also Gemini's free model.
2.0 Flash 2025-01-30.11.2026 Old version, no longer maintained: Discontinued Developed by Google with a focus on multimodality, agentic capabilities, and speed.[81]
2.0 Flash-Lite 2025-02-25.11.2026 Old version, no longer maintained: Discontinued First-ever Gemini Flash-Lite model designed for cost-efficiency and speed.[82]
2.0 Pro 2025-02-05.11.2026 Old version, no longer maintained: Discontinued
2.5 Pro 2025-03-25.11.2026 Joriy versiya: Latest version
2.5 Flash 2025-04-17.11.2026 Eski versiya (qoʻllab-quvvatlanadi): Supported An incremental improvement from Gemini 2.5.
2.5 Flash-Lite 2025-06-17.11.2026 Eski versiya (qoʻllab-quvvatlanadi): Supported
2.5 Flash Image (Nano Banana) 2025-08-26.11.2026 (Publicly) Eski versiya (qoʻllab-quvvatlanadi): Supported
3 Pro 2025-11-18.11.2026 Old version, no longer maintained: Discontinued preview[83] Gemini 3 Pro is not a modification or a fine-tune or a prior model. It is the foundational model of the Gemini 3 series. A sparse mixture-of-experts, natively multimodal, reasoning model. It can process text, visual, and audio inputs. It supports up to 1,000,000 input tokens and 64,000 output tokens.[84]
3 Pro Image (Nano Banana Pro) 2025-11-20.11.2026 Joriy versiya: Latest version An improved version of Nano Banana, which includes better text rendering and better real world knowledge. It was released as a preview model on 2025-11-20.11.2026, and made generally available on 2026-05-28.11.2026.[85]
3 Deep Think 2025-12-03.11.2026 Kelgusidagi reliz: Preview Based on the 2.5 Pro "Deep Think" mode that achieved high ratings in IOI[86][87]
3 Flash 2025-12-17.11.2026 Kelgusidagi reliz: Preview Based on Gemini 3 Pro.
3.1 Pro 2026-02-19.11.2026 Kelgusidagi reliz: Preview Based on Gemini 3 Pro. Created to handle complex reasoning tasks across large datasets and diverse inputs, including text, audio, image, video, and entire codebases.[88]
3.1 Flash Image

(Nano Banana 2)

2026-02-26.11.2026 Joriy versiya: Latest version Based on Gemini 3 Flash. Designed to generate both image and text outputs.[89]
3.1 Flash-Lite 2026-03-03.11.2026 Eski versiya (qoʻllab-quvvatlanadi): Supported Based on Gemini 3 Pro. It is optimized for high-volume, latency-sensitive tasks such as translation, and text classification.[90]
3.5 Flash 2026-05-19.11.2026 Eski versiya (qoʻllab-quvvatlanadi): Supported Based on Gemini 3 Flash.[91]
3.1 Flash-Lite Image

(Nano Banana 2 Lite)

2026-06-30.11.2026 Joriy versiya: Latest version Based on Gemini 3.1 Flash-Lite. It is optimized for cost-effective image generation.[92]
3.5 Flash-Lite 2026-07-21.11.2026 Joriy versiya: Latest version Based on Gemini 3.1 Flash-Lite. It is optimized for high-volume, latency-sensitive tasks such as translation, text classification, and supporting agentic workflows.[93]
3.6 Flash 2026-07-21.11.2026 Eski versiya (qoʻllab-quvvatlanadi): Supported Based on Gemini 3.5 Flash. Offers improvements in coding, general knowledge, and multimodal processing, alongside increased token efficiency relative to its predecessor.[94]
3.7 Flash 2026-08-13.11.2026 Eski versiya (qoʻllab-quvvatlanadi): Supported Based on Gemini 3.6 Flash. It features improvements to its core reasoning foundation.[95]
3.8 Flash 2026-09-02.11.2026 Joriy versiya: Latest version Based on Gemini 3.7 Flash.[96] A version with a more permissive set of mitigations for cybersecurity, 3.8 Flash Cyber, is available through Google's Fairwind Program.

Nano Banana

[tahrir | manbasini tahrirlash]
"Google Pics" redirects here; not to be confused with Google Photos yoki Google Images.
Yana qarang: AI content watermarking
AI-generated abstract art created with Nano Banana 2 at Gemini
Geminida Nano Banana 2 yordamida yaratilgan sunʼiy intellekt abstract art

Nano Banana (rasmiy nomi Gemini 2.5 Flash Image), Nano Banana Pro (rasmiy nomi Gemini 3 Pro Image) va Nano Banana 2 (rasmiy nomi Gemini 3.1 Flash Image) — tasvir yaratish va tahrirlash modellari.

«Nano Banana» — bu modelning Arena platformasida yashirin ommaviy sinovdan oʻtayotgan paytdagi kod nomi edi. U birinchi marta 2025-yil 12-avgustda Arena nomli sunʼiy intellekt baholash platformasida anonim model sifatida paydo boʻldi. Model 2025-yil 26-avgustda Gemini ilovasi va Google AI xizmatlari orqali ommaga taqdim etildi. «Nano Banana» laqabi Google DeepMind mahsulot menejeri Naina Raisinghaniga berilgan laqablardan kelib chiqqan.[97] Keyinchalik Google rasmiy eʼlonida uning asl nomi Gemini 2.5 Flash Image ekanini tasdiqladi.[98][99] 2025-yil 20-noyabrda DeepMind Nano Banana Pro (Gemini 3 Pro Image) modelini chiqardi. Unda matn tasvirlash va dunyo haqidagi bilim sezilarli yaxshilandi.[85][100][101][99]

Chiqqanidan soʻng Nano Banana ijtimoiy tarmoqlarda virusli internet hodisasiga aylandi. Ayniqsa, uning fotorealistik «3D haykalcha» tasvirlari katta shuhrat qozondi. Model Gemini ilovasi, Google AI Studio va Vertex AI orqali foydalanuvchilarga taqdim etildi. Googlening maʼlum qilishicha, u Gemini ilovasiga 10 milliondan ortiq yangi foydalanuvchi jalb qilgan. Shuningdek, ishga tushirilganidan bir necha hafta ichida 200 milliondan ziyod tasvir tahriri amalga oshirilgan.[102][103]

Model foydalanuvchilarga soch turmagi va fonni oʻzgartirish hamda tabiiy tildagi koʻrsatmalar yordamida suratlarni birlashtirish imkonini beradi. Obyekt izchilligi tufayli bir xil shaxs yoki narsa tasvirning turli tahrirlarida ham tanib olinadi. Koʻp tasvirni birlashtirish funksiyasi bir necha fotosuratni yaxlit bir natijaga aylantiradi, dunyo haqidagi bilim esa kontekstga mos oʻzgarishlar kiritishga yordam beradi. Model, shuningdek, SynthID raqamli belgisini qoʻyadi. Bu — natijalardagi koʻrinmas raqamli imzo boʻlib, sunʼiy intellekt yaratgan maʼlumotni aniqlashga xizmat qiladi.[99][104] Koʻp tasvirni birlashtirish funksiyasi bir necha fotosuratni yaxlit bir natijaga aylantiradi, dunyo haqidagi bilim esa kontekstga mos oʻzgarishlar kiritishga yordam beradi. Odamlar Nano Bananani virusli mashhur boʻlgan bir moda bilan bogʻlay boshladi: unda foydalanuvchilar oʻz selfilarini oʻyinchoqqa oʻxshash 3D haykalchalarga aylantirgan. Bu hodisa Instagram va X (avvalgi Twitter) kabi saytlarda tez tarqaldi.[105][106] Model X tarmogʻiga qoʻshilgach, foydalanuvchilar postlarda Nano Bananani bevosita belgilab, koʻrsatmalar asosida surat yarata olishdi. Bu esa uning shuhratini yanada oshirdi.[105]

TechRadarning 2025-yil sentabrdagi sharhida Nano Banana turli koʻrsatmalarda ChatGPTning tasvir yaratish xizmatiga qaraganda realroq va izchilroq ekani qayd etildi.[107] Tom's Guide sharhida uning ijodiy va jonli tasvir tahrirlarini bajara olishi yuqori baholandi.[108] PC Gamer sharhida esa modelda kesish kabi ayrim oddiy tahrir vositalari yoʻqligi taʼkidlandi. Bundan tashqari, mahsulot baʼzan oʻzgarishlarni qoʻllamay, asl tasvirni qaytargani ham qayd etildi.[104] Nano Banana meʼmoriy vizualizatsiya sohasida yaxshi natija koʻrsatdi: u murakkab geometriyada ham tasvirlarni toʻgʻri miqyosda yaratdi.[109][104]

2026-yil 26-fevralda Nano Banana 2 ishga tushirildi hamda Gemini chatboti, Search AI Mode va Lens xizmatlariga integratsiya qilindi. Bu Gemini 3.1 Flash Image asosidagi tezroq versiya boʻlib, koʻrsatmalarga amal qilish va matn tasvirlash imkoniyatlari yaxshilangan.[66] May oyida boʻlib oʻtgan 2026-yilgi Google I/O konferensiyasida Google Nano Banana asosidagi Google Pics nomli tasvir yaratish platformasini eʼlon qildi. U Google Workspace toʻplamining bir qismi hisoblanadi.[110]

2026-yil 30-iyulda Google Nano Bananani Google Earth bilan integratsiya qildi. Ammo bu xizmat 24 soat ichida oʻchirildi. Sababi — tadqiqotchilar, ochiq manba tekshiruvchilari va ijtimoiy tarmoq foydalanuvchilarining noroziligi boʻldi, chunki xizmat notoʻgʻri maʼlumot tarqatishga moyil edi. Foydalanuvchilar koʻrsatmalarda cheklovlar yoʻqligini isbotlash uchun yirik falokatlar va harbiy harakatlarning soxta tasvirlarini yaratdi. Ular orasida eng koʻzga tashlangani 11-sentabr terror hujumlari manzarasini qayta yaratish boʻldi.[111][112][113]

Texnik tavsiflar

[tahrir | manbasini tahrirlash]

Gemini multimodal boʻlgani uchun har bir kontekst oynasi bir necha xil kirish maʼlumotini oʻz ichiga olishi mumkin. Turli rejimlarni aralashtirib qoʻllash mumkin, ularni qatʼiy tartibda berish shart emas. Bu multimodal suhbat oʻtkazishga imkon beradi. Masalan, foydalanuvchi suhbatni matn, rasm, video va audio aralashmasi bilan istalgan tartibda boshlashi mumkin. Gemini ham xuddi shunday erkin tartibda javob berishi mumkin. Kiritiladigan tasvirlar turli oʻlchamlarda boʻlishi mumkin, video esa tasvirlar ketma-ketligi sifatida kiritiladi. Audio 16 kHz chastotada namunalanadi, soʻngra Universal Speech Model uni tokenlar ketma-ketligiga aylantiradi. Gemininig maʼlumotlar toʻplami multimodal va koʻp tilli boʻlib, u «veb-hujjatlar, kitoblar va kod hamda tasvir, audio va video maʼlumotlarni» oʻz ichiga oladi.[114]

Gemini va Gemma modellari faqat dekoderdan iborat transformerlardir. Ularga TPU qurilmalarida samarali oʻqitish va xulosa chiqarish uchun oʻzgartirishlar kiritilgan. 1.0 avlodi koʻp soʻrovli eʼtibor mexanizmidan foydalanadi.[114] Biroq Gemini 2.0, 2.5 va 3 uchun oq kitoblar chop etilmagan.

Technical specifications of Gemini models
Generation Variant Release date Parameters Context length Notes
1.0 Nano-1 2023-12-06.11.2026 1.8B 32,768 Distilled from "larger Gemini models", 4-bit quantized[114]
Nano-2 2023-12-06.11.2026 3.25B
Pro 2023-12-13.11.2026 Not publicly known
Ultra 2024-02-08.11.2026 Not publicly known
1.5 Pro 2024-02-15.11.2026 Not publicly known 10,000,000[115][116] 1 million tokens in production API
Mini 2024-05-14.11.2026 Not publicly known

Baholanishi

[tahrir | manbasini tahrirlash]
Yana qarang: Google Gemini#Reception

MIT Technology Review Gemini taqdimoti atrofidagi mish-mishlarni «sunʼiy intellekt shov-shuvining choʻqqisi» deb taʼrifladi.[117][22] Business Insiderdan Hugh Langley Gemini Google uchun hal qiluvchi lahza ekanini yozdi: «Agar Gemini hayratga solsa, bu Googlega Microsoft va OpenAI uni ogohlantirmay ortda qoldirdi degan fikrni oʻzgartirishga yordam beradi. Agar u umidni oqlamasa, Google ortda qoldi deydigan tanqidchilarni yanada dadillashtiradi».[118]

2023-yil dekabrdagi taqdimotga munosabat bildirgan informatika olimi Oren Etzioni Google Deepmind va OpenAI oʻrtasida qurollanish poygasi boshlanishini bashorat qildi. Informatika olimi Alexei Efros Gemininig multimodal yondashuvi salohiyatini yuqori baholadi.[21] Professorlar Percy Liang, Emily Bender va Michael Madden ogohlantirdi: oʻqitishda ishlatilgan maʼlumotlarni bilmay turib, benchmark natijalarini talqin qilish qiyin.[117][119] Fast Company uchun yozgan Mark Sullivan Googlening iPhone bozordagi ustunligiga raqobat qilish imkoniyati borligini taʼkidladi. Uning fikricha, Apple oʻzining Siri virtual yordamchisi bilan Geminiga oʻxshash imkoniyatni yarata olishi dargumon edi.[120] Gemini taqdimotidan keyingi kuni Google aksiyalari 5,3 foizga koʻtarildi.[121][122]

Google Gemini haqidagi namoyish videosi uchun tanqidga uchradi. Bu video modelning real vaqtdagi muloqot imkoniyatlarini haqiqiy koʻrsatuv sifatida taqdim etilgan edi, aslida esa shunday koʻrinishi uchun tahrir qilingandi. Shunga qaramay, Google boshqa joylarda (video tashqarisida) kichik izohlar berib, uning aslida real vaqtda yozib olinmaganini ogohlantirgan edi.[123]

Gemini 2.5 Pro Experimental LMArena reytingida birinchi oʻrinda paydo boʻldi; bu reyting odamlarning afzal koʻrishini oʻlchaydi.[49][51] Model turli sunʼiy intellekt benchmarklarida, jumladan Humanity's Last Exam sinovida yuqori natijalarga erishdi.[49][124][51][50] Dastlabki sharhlarda uning yaxshilangan mulohaza yuritish qobiliyati va oldingi versiyalarga nisbatan ustunligi qayd etildi.[50][52] Eʼlon qilingan benchmarklar, shuningdek, Anthropic, xAI yoki OpenAI kabi raqobatchilarning oʻsha davrdagi modellari ustun boʻlgan sohalarni ham koʻrsatdi.[124][51]

Yana qarang

[tahrir | manbasini tahrirlash]

Manbalar

[tahrir | manbasini tahrirlash]
  1. ↑ 1,0 1,1 „Gemini 3 Flash: Frontier intelligence built for speed“ (2025-yil 17-dekabr).
  2. ↑ 2,0 2,1 „Gemini 3.1 Pro: A smarter model for your most complex tasks“ (en-us). Google (2026-yil 19-fevral). Qaraldi: 2026-yil 19-fevral.
  3. ↑ Grant, Nico. „Google Builds on Tech's Latest Craze With Its Own A.I. Products“ (2023-yil 10-may). 2023-yil 10-mayda asl nusxadan arxivlangan. Qaraldi: 2023-yil 21-avgust.
  4. ↑ Ortiz, Sabrina „Every major AI feature announced at Google I/O 2023“. ZDNet (2023-yil 10-may). 2023-yil 10-mayda asl nusxadan arxivlangan. Qaraldi: 2023-yil 21-avgust.
  5. ↑ 5,0 5,1 5,2 5,3 Milmo, Dan. „Google says new AI model Gemini outperforms ChatGPT in most tests“ (2023-yil 6-dekabr). 2023-yil 6-dekabrda asl nusxadan arxivlangan. Qaraldi: 2023-yil 6-dekabr.
  6. ↑ Levy, Steven (September 11, 2023). „Sundar Pichai on Google;s AI, Microsoft's AI, OpenAI, and ... Did We Mention AI?“. Wired. September 11, 2023da asl nusxadan arxivlandi. Qaraldi: September 12, 2023.
  7. ↑ Knight, Will (June 26, 2023). „Google DeepMind's CEO Says Its Next Algorithm Will Eclipse ChatGPT“. Wired. June 26, 2023da asl nusxadan arxivlandi. Qaraldi: August 21, 2023.
  8. ↑ 8,0 8,1 8,2 Victor, Jon „How Google is Planning to Beat OpenAI“. The Information (2023-yil 15-avgust). 2023-yil 15-avgustda asl nusxadan arxivlangan. Qaraldi: 2023-yil 21-avgust.
  9. ↑ Grant, Nico. „Google Calls In Help From Larry Page and Sergey Brin for A.I. Fight“ (2023-yil 20-yanvar). 2023-yil 20-yanvarda asl nusxadan arxivlangan. Qaraldi: 2023-yil 6-fevral.
  10. ↑ Kruppa, Miles; Seetharaman, Deepa. „Sergey Brin Is Back in the Trenches at Google“ (2023-yil 21-iyul). 2023-yil 21-iyulda asl nusxadan arxivlangan. Qaraldi: 2023-yil 7-sentyabr.
  11. ↑ Carter, Tom „Google confirms that its cofounder Sergey Brin played a key role in creating its ChatGPT rival“. Business Insider (2023-yil 7-dekabr). 2023-yil 7-dekabrda asl nusxadan arxivlangan. Qaraldi: 2023-yil 31-dekabr.
  12. ↑ Victor, Jon „OpenAI Hustles to Beat Google to Launch 'Multimodal' LLM“. The Information (2023-yil 18-sentyabr). 2023-yil 18-sentyabrda asl nusxadan arxivlangan. Qaraldi: 2023-yil 15-oktyabr.
  13. ↑ „Google nears release of AI software Gemini, The Information reports“. Reuters (2023-yil 14-sentyabr). 2023-yil 15-sentyabrda asl nusxadan arxivlangan. Qaraldi: 2023-yil 2-oktyabr.
  14. ↑ Nolan, Beatrice „Google is quietly handing out early demos of its GPT-4 rival called Gemini. Here's what we know so far about the upcoming AI model.“. Business Insider (2023-yil 23-sentyabr). 2023-yil 23-sentyabrda asl nusxadan arxivlangan. Qaraldi: 2023-yil 16-oktyabr.
  15. ↑ 15,0 15,1 15,2 15,3 Kruppa, Miles. „Google Announces AI System Gemini After Turmoil at Rival OpenAI“ (2023-yil 6-dekabr). 2023-yil 6-dekabrda asl nusxadan arxivlangan. Qaraldi: 2023-yil 6-dekabr.
  16. ↑ 16,0 16,1 16,2 16,3 Liedtike, Michael; O'Brien, Matt. „Google launches Gemini, upping the stakes in the global AI race“. Associated Press (2023-yil 6-dekabr). 2023-yil 6-dekabrda asl nusxadan arxivlangan. Qaraldi: 2023-yil 6-dekabr.
  17. ↑ Edwards, Benj „Google launches Gemini—a powerful AI model it says can surpass GPT-4“. Ars Technica (2023-yil 6-dekabr). 2023-yil 6-dekabrda asl nusxadan arxivlangan. Qaraldi: 2023-yil 6-dekabr.
  18. ↑ 18,0 18,1 Pierce, David „Google launches Gemini, the AI model it hopes will take down GPT-4“. The Verge (2023-yil 6-dekabr). 2023-yil 6-dekabrda asl nusxadan arxivlangan. Qaraldi: 2023-yil 6-dekabr.
  19. ↑ Fung, Brian; Thorbecke, Catherine „Google launches Gemini, its most-advanced AI model yet, as it races to compete with ChatGPT“. CNN Business (2023-yil 6-dekabr). 2023-yil 6-dekabrda asl nusxadan arxivlangan. Qaraldi: 2023-yil 6-dekabr.
  20. ↑ „Google launches Gemini, upping the stakes in the global AI race“. CBS News (2023-yil 6-dekabr). 2023-yil 7-dekabrda asl nusxadan arxivlangan. Qaraldi: 2023-yil 7-dekabr.
  21. ↑ 21,0 21,1 21,2 Knight, Will (December 6, 2023). „Google Just Launched Gemini, Its Long-Awaited Answer to ChatGPT“. Wired. December 6, 2023da asl nusxadan arxivlandi. Qaraldi: December 6, 2023.
  22. ↑ 22,0 22,1 Henshall, Will „Google DeepMind Unveils Its Most Powerful AI Offering Yet“ (2023-yil 6-dekabr). 2023-yil 6-dekabrda asl nusxadan arxivlangan. Qaraldi: 2023-yil 6-dekabr.
  23. ↑ Metz, Cade; Grant, Nico. „Google Updates Bard Chatbot With 'Gemini' A.I. as It Chases ChatGPT“ (2023-yil 6-dekabr). 2023-yil 6-dekabrda asl nusxadan arxivlangan. Qaraldi: 2023-yil 6-dekabr.
  24. ↑ Elias, Jennifer „Google launches its largest and 'most capable' AI model, Gemini“. CNBC (2023-yil 6-dekabr). 2023-yil 6-dekabrda asl nusxadan arxivlangan. Qaraldi: 2023-yil 6-dekabr.
  25. ↑ Alba, Davey; Ghaffary, Shirin. „Google Opens Access to Gemini, Racing to Catch Up to OpenAI“. Bloomberg News (2023-yil 6-dekabr). 2023-yil 6-dekabrda asl nusxadan arxivlangan. Qaraldi: 2023-yil 7-dekabr.
  26. ↑ Knight, Will (December 6, 2023). „Google DeepMind's Demis Hassabis Says Gemini Is a New Breed of AI“. Wired. December 6, 2023da asl nusxadan arxivlandi. Qaraldi: December 7, 2023.
  27. ↑ Google announces Gemini CLI: your open-source AI agent. Blog.google. Retrieved 28 August 2025.
  28. ↑ Gurman, Mark; Love, Julia; Alba, Davey. „Samsung Bets on Google-Powered AI Features in Smartphone Revamp“. Bloomberg News (2024-yil 17-yanvar). 2024-yil 17-yanvarda asl nusxadan arxivlangan. Qaraldi: 2024-yil 4-fevral.
  29. ↑ Chokkattu, Julian (January 17, 2024). „Samsung's Galaxy S24 Phones Call on Google's AI to Spruce Up Their Smarts“. Wired. January 17, 2024da asl nusxadan arxivlandi. Qaraldi: February 4, 2024.
  30. ↑ Metz, Cade. „Google Releases Gemini, an A.I.-Driven Chatbot and Voice Assistant“ (2024-yil 8-fevral). 2024-yil 8-fevralda asl nusxadan arxivlangan. Qaraldi: 2024-yil 8-fevral.
  31. ↑ Dastin, Jeffrey. „Google rebrands Bard chatbot as Gemini, rolls out paid subscription“. Reuters (2024-yil 8-fevral). 2024-yil 8-fevralda asl nusxadan arxivlangan. Qaraldi: 2024-yil 8-fevral.
  32. ↑ Li, Abner „Google One AI Premium is $19.99/mo with Gemini Advanced & Gemini for Workspace“. 9to5Google (2024-yil 8-fevral). 2024-yil 8-fevralda asl nusxadan arxivlangan. Qaraldi: 2024-yil 8-fevral.
  33. ↑ Mehta, Ivan „Google's Bard chatbot gets the Gemini Pro update globally“. TechCrunch (2024-yil 1-fevral). 2024-yil 1-fevralda asl nusxadan arxivlangan. Qaraldi: 2024-yil 4-fevral.
  34. ↑ Knight, Will (February 15, 2024). „Google's Flagship AI Model Gets a Mighty Fast Upgrade“. Wired. February 15, 2024da asl nusxadan arxivlandi. Qaraldi: February 21, 2024.
  35. ↑ Nieva, Richard (February 15, 2024). „Google Unveils Gemini 1.5, But Only Developers And Enterprise Clients Have Access For Now“. Forbes. February 15, 2024da asl nusxadan arxivlandi. Qaraldi: February 21, 2024.
  36. ↑ McCracken, Harry (February 15, 2024). „Google's new Gemini 1.5 AI can dive deep into oceans of video and audio“. Fast Company. February 17, 2024da asl nusxadan arxivlandi. Qaraldi: February 21, 2024.
  37. ↑ Stokes, Samantha „Here's everything you need to know about Gemini 1.5, Google's newly updated AI model that hopes to challenge OpenAI“. Business Insider (2024-yil 15-fevral). 2024-yil 19-fevralda asl nusxadan arxivlangan. Qaraldi: 2024-yil 21-fevral.
  38. ↑ Khan, Jeremy (February 21, 2024). „Google unveils new family of open-source AI models called Gemma to take on Meta and others—deciding open-source AI ain't so bad after all“. Fast Company. February 21, 2024da asl nusxadan arxivlandi. Qaraldi: February 21, 2024.
  39. ↑ Alba, Davey. „Google Delves Deeper Into Open Source with Launch of Gemma AI Model“. Bloomberg News (2024-yil 21-fevral). 2024-yil 21-fevralda asl nusxadan arxivlangan. Qaraldi: 2024-yil 21-fevral.
  40. ↑ Metz, Cade; Grant, Nico. „Google Is Giving Away Some of the A.I. That Powers Chatbots“ (2024-yil 21-fevral). 2024-yil 21-fevralda asl nusxadan arxivlangan. Qaraldi: 2024-yil 21-fevral.
  41. ↑ Elias, Jennifer „Google rolls out its most powerful AI models as competition from OpenAI heats up“. CNBC (2024-yil 12-avgust). 2024-yil 14-mayda asl nusxadan arxivlangan. Qaraldi: 2024-yil 13-avgust.
  42. ↑ „Updated production-ready Gemini models, reduced 1.5 Pro pricing, increased rate limits, and more- Google Developers Blog“ (en). developers.googleblog.com.
  43. ↑ „Introducing Gemini 2.0: our new AI model for the agentic era“ (en-us). Google (2024-yil 11-dekabr). Qaraldi: 2024-yil 20-dekabr.
  44. ↑ „The next chapter of the Gemini era for developers- Google Developers Blog“ (en). developers.googleblog.com. Qaraldi: 2024-yil 20-dekabr.
  45. ↑ „Gemini 2.0 Flash (experimental) | Gemini API“ (en). Google AI for Developers. Qaraldi: 2024-yil 20-dekabr.
  46. ↑ „Gemini 2.0 is now available to everyone“ (en-us). Google (2025-yil 5-fevral).
  47. ↑ Parada, Carolina „Gemini Robotics brings AI into the physical world“. Google DeepMind (2025-yil 12-mart). Qaraldi: 2025-yil 11-iyul.
  48. ↑ Abner, Li „Gemini in Android Studio can now turn UI mockups into code“. 9to5Google (2025-yil 13-mart). Qaraldi: 2025-yil 12-aprel.
  49. ↑ 49,0 49,1 49,2 49,3 Kavukcuoglu, Koray „Gemini 2.5: Our most intelligent AI model“. The Keyword (Google Blog) (2025-yil 25-mart). Qaraldi: 2025-yil 26-mart.
  50. ↑ 50,0 50,1 50,2 „Gemini 2.5 is now available for Advanced users and it seriously improves Google's AI reasoning“. TechRadar (2025-yil 26-mart). Qaraldi: 2025-yil 26-mart.
  51. ↑ 51,0 51,1 51,2 51,3 51,4 „Google's Gemini 2.5 Pro model tops LMArena by close to 40 points“. R&D World (2025-yil 25-mart). Qaraldi: 2025-yil 26-mart.
  52. ↑ 52,0 52,1 „Gemini 2.5 Pro Exp: How to Access, Features, Applications & More“. Analytics Vidhya (2025-yil 26-mart). Qaraldi: 2025-yil 26-mart.
  53. ↑ „Gemini 2.5: Our most intelligent models are getting even better“ (2025-yil 20-may). Qaraldi: 2025-yil 21-may.
  54. ↑ „Deep Think boosts the performance of Google's flagship Gemini AI model“. TechCrunch (2025-yil 20-may). Qaraldi: 2025-yil 21-may.
  55. ↑ 55,0 55,1 Doshi, Tulsee „We're expanding our Gemini 2.5 family of models“. The Keyword. Google (2025-yil 17-iyun). Qaraldi: 2025-yil 12-iyul.
  56. ↑ „A new era of intelligence with Gemini 3“ (en-us). Google (2025-yil 18-noyabr). 2025-yil 29-noyabrda asl nusxadan arxivlangan. Qaraldi: 2025-yil 18-noyabr.
  57. ↑ Warren, Tom „OpenAI's GPT-5.2 'code red' response to Google is coming next week“. The Verge (2025-yil 5-dekabr). Qaraldi: 2025-yil 9-dekabr.
  58. ↑ Hart, Robert „OpenAI declares 'code red' as Google catches up in AI race“. The Verge (2025-yil 2-dekabr). Qaraldi: 2025-yil 9-dekabr.
  59. ↑ Edwards, Benj „OpenAI releases GPT-5.2 after "code red" Google threat alert“. Ars Technica (2025-yil 11-dekabr). Qaraldi: 2025-yil 15-dekabr.
  60. ↑ Bellan, Rebecca „OpenAI fires back at Google with GPT-5.2 after 'code red' memo“. TechCrunch (2025-yil 11-dekabr). Qaraldi: 2025-yil 15-dekabr.
  61. ↑ „Gemini 3 Deep Think is now available in the Gemini app.“ (en-us). Google (2025-yil 4-dekabr). Qaraldi: 2025-yil 5-dekabr.
  62. ↑ „Gemini 3 Flash: frontier intelligence built for speed“ (en-us). Google (2025-yil 17-dekabr). Qaraldi: 2025-yil 17-dekabr.
  63. ↑ Duffy, Clare. „Apple teams up with Google Gemini for AI-powered Siri“ (en). CNN (2026-yil 12-yanvar). Qaraldi: 2026-yil 15-yanvar.
  64. ↑ „Fashion’s biggest brands leaning into AI“ (en). NBC News. Qaraldi: 2026-yil 27-aprel.
  65. ↑ Klinken, Erik „Google launches Gemini 3.1 Pro, an LLM for complex reasoning“. Techzine (2026-yil 20-fevral). 2026-yil 28-fevralda asl nusxadan arxivlangan. Qaraldi: 2026-yil 21-aprel.
  66. ↑ 66,0 66,1 Murti, Lola „Google launches Nano Banana 2, updating its viral AI image generator“ (en). CNBC (2026-yil 26-fevral). Qaraldi: 2026-yil 28-fevral.
  67. ↑ „Gemini 3.1 Flash-Lite: Built for intelligence at scale“ (en-us). Google (2026-yil 3-mart). Qaraldi: 2026-yil 3-mart.
  68. ↑ Kasanmascheff, Markus „Google Launches Gemini 3.1 Flash-Lite for Enterprise Scale“ (en-US). WinBuzzer (2026-yil 3-mart). Qaraldi: 2026-yil 3-mart.
  69. ↑ https://blog.google/innovation-and-ai/technology/developers-tools/gemma-4/
  70. ↑ „Gemini 3.5: frontier intelligence with action“ (en-us). Google (2026-yil 19-may). Qaraldi: 2026-yil 17-iyun.
  71. ↑ „Introducing Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber“ (en-us). Google (2026-yil 21-iyul). Qaraldi: 2026-yil 21-iyul.
  72. ↑ Sigalos, MacKenzie „Google expands Gemini lineup with cheaper models and new Mythos rival“ (en). CNBC (2026-yil 21-iyul). Qaraldi: 2026-yil 21-iyul.
  73. ↑ „Introducing Gemini 3.7 Flash“ (en-us). Google (2026-yil 13-avgust). Qaraldi: 2026-yil 13-avgust.
  74. ↑ „Introducing Gemini 3.8 Flash and 3.8 Flash Cyber“ (en-us). Google (2026-yil 2-sentyabr). Qaraldi: 2026-yil 6-sentyabr.
  75. ↑ „Gemini Release updates“. Google. 2025-yil 9-aprelda asl nusxadan arxivlangan. Qaraldi: 2025-yil 9-aprel.
  76. ↑ „Gemini models“. Google. 2025-yil 9-aprelda asl nusxadan arxivlangan. Qaraldi: 2025-yil 9-aprel.
  77. ↑ „Gemini deprecations | Gemini API | Google AI for Developers“ (en). Google AI for Developers. 2026-yil 26-mayda asl nusxadan arxivlangan. Qaraldi: 2026-yil 27-may.
  78. ↑ „Model versions and lifecycle“. Google Cloud.
  79. ↑ 79,0 79,1 79,2 Pichai, Sundar; Hassabis, Demis „Introducing Gemini: our largest and most capable AI model“. The Keyword (2023-yil 6-dekabr). Qaraldi: 2025-yil 12-iyul.
  80. ↑ Pichai, Sundar; Hassabis, Demiss „Our next-generation model: Gemini 1.5“. The Keyword (2024-yil 15-fevral). Qaraldi: 2025-yil 12-iyul.
  81. ↑ Pichai, Sundar; Hassabis, Demis; Kavukcuoglu, Koray „Introducing Gemini 2.0: our new AI model for the agentic era“. The Keyword (2024-yil 11-dekabr). Qaraldi: 2025-yil 12-iyul.
  82. ↑ Mallick, Shrestha; Kilpatrick, Logan „Gemini 2.0: Flash, Flash-Lite and Pro“. Google for Developers. Qaraldi: 2025-yil 12-iyul.
  83. ↑ https://discuss.ai.google.dev/t/migrate-from-gemini-3-pro-preview-to-gemini-3-1-pro-preview-before-march-9-2026/127062
  84. ↑ [Gemini 3 Pro] External Model Card - November 18, 2025 - v5
  85. ↑ 85,0 85,1 „Introducing Nano Banana Pro“ (en-us). Google (2025-yil 20-noyabr). Qaraldi: 2025-yil 20-noyabr.
  86. ↑ „Gemini 3 Deep Think is now available in the Gemini app“ (2025-yil 4-dekabr).
  87. ↑ „Advanced version of Gemini with Deep Think officially achieves gold-medal standard at the International Mathematical Olympiad“.
  88. ↑ „Gemini 3.1 Pro - Model Card“ (en). Google DeepMind. Qaraldi: 2026-yil 23-iyul.
  89. ↑ „Gemini 3.1 Flash Image - Model Card“ (en). Google DeepMind. Qaraldi: 2026-yil 23-iyul.
  90. ↑ „Gemini 3.1 Flash-Lite - Model Card“ (en). Google DeepMind. Qaraldi: 2026-yil 23-iyul.
  91. ↑ Bellan, Rebecca „With Gemini 3.5 Flash, Google bets its next AI wave on agents, not chatbots“ (en-US). TechCrunch (2026-yil 19-may). Qaraldi: 2026-yil 19-may.
  92. ↑ „Gemini 3.1 Flash-Lite Image - Model Card“ (en). Google DeepMind. Qaraldi: 2026-yil 30-iyul.
  93. ↑ „Gemini 3.5 Flash-Lite - Model Card“ (en). Google DeepMind. Qaraldi: 2026-yil 23-iyul.
  94. ↑ „Gemini 3.6 Flash - Model Card“ (en). Google DeepMind. Qaraldi: 2026-yil 23-iyul.
  95. ↑ „Gemini 3.7 Flash - Model Card“ (en). Google DeepMind. Qaraldi: 2026-yil 13-avgust.
  96. ↑ „Gemini 3.8 Flash - Model Card“ (en). Google DeepMind. Qaraldi: 2026-yil 2-sentyabr.
  97. ↑ Griffiths, Brent D. „She named Google's Nano Banana. She has an idea why it went viral.“ (en-US). Business Insider. Qaraldi: 2026-yil 20-yanvar.
  98. ↑ País, El „Google se vuelve viral con Nano Banana, su modelo más avanzado de edición de imágenes con IA“ (es). El País (2025-yil 29-avgust). Qaraldi: 2025-yil 10-sentyabr.
  99. ↑ 99,0 99,1 99,2 „Image editing in Gemini just got a major upgrade“ (en-us). Google (2025-yil 26-avgust). Qaraldi: 2025-yil 10-sentyabr.
  100. ↑ „Build with Nano Banana Pro, our Gemini 3 Pro Image model“ (en-us). Google (2025-yil 20-noyabr). Qaraldi: 2025-yil 20-noyabr.
  101. ↑ „7 tips to get the most out of Nano Banana Pro“ (en-us). Google (2025-yil 20-noyabr). Qaraldi: 2025-yil 20-noyabr.
  102. ↑ Li, Abner „Nano Banana responsible for 10+ million first-time Gemini app users“ (en-US). 9to5Google (2025-yil 4-sentyabr). Qaraldi: 2025-yil 10-sentyabr.
  103. ↑ „Google celebrates major achievements by its Gemini app 'Nano Banana' editor“ (en). Android Central (2025-yil 5-sentyabr). Qaraldi: 2025-yil 10-sentyabr.
  104. ↑ 104,0 104,1 104,2 Kinghorn, Jess. „Gemini's 'Nano Banana' AI image editor can't crop a picture, but its penchant for deepfakes 'while keeping you, you' makes me want to wear a brown paper bag on my head forever more“ (en). PC Gamer (2025-yil 28-avgust). Qaraldi: 2025-yil 10-sentyabr.
  105. ↑ 105,0 105,1 „Google's Nano Banana available on X: Here's how to use“. The Economic Times (2025-yil 6-sentyabr). Qaraldi: 2025-yil 10-sentyabr.
  106. ↑ „Nano Banana' trend: What is the viral 3D figurine craze, and how to create one for free“. The Times of India (2025-yil 10-sentyabr). Qaraldi: 2025-yil 10-sentyabr.
  107. ↑ „Gemini's Nano Banana makes photo editing fun again – and yes, it's better than ChatGPT“ (en). TechRadar (2025-yil 1-sentyabr). Qaraldi: 2025-yil 10-sentyabr.
  108. ↑ Caswell, Amanda „Nano Banana just broke the internet with these viral trends — I tried these 5 prompts and I'm blown away“ (en). Tom's Guide (2025-yil 9-sentyabr). Qaraldi: 2025-yil 10-sentyabr.
  109. ↑ Nelson, Michael „Nano-Banana and Architectural Visualization for Adaptive Reuse Projects“ (en-US). AMICO Architectural Metals (2025-yil 11-sentyabr). Qaraldi: 2025-yil 11-sentyabr.
  110. ↑ Peters, Jay „Google Pics is a new app that tries to fix AI image editing“. The Verge (2026-yil 19-may). 2026-yil 19-mayda asl nusxadan arxivlangan. Qaraldi: 2026-yil 19-may.
  111. ↑ Gibbs, Alice. „Google Earth pulls “Nano Banana” after users create offensive images“ (en). Newsweek (2026-yil 3-avgust). Qaraldi: 2026-yil 3-avgust.
  112. ↑ Hassan, Adeel. „Google Earth Disables A.I. Tool After One Day Over Disinformation Concerns“. The New York Times (2026-yil 2-avgust). Qaraldi: 2026-yil 3-avgust.
  113. ↑ Wong, Matteo. „Google Just Ruined One of Its Most Important Tools“ (en). The Atlantic (2026-yil 31-iyul). Qaraldi: 2026-yil 3-avgust.
  114. ↑ 114,0 114,1 114,2 Gemini: A Family of Highly Capable Multimodal Models (PDF) (Technical report). Google DeepMind. December 6, 2023. Archived (PDF) from the original on December 6, 2023. Retrieved December 7, 2023.
  115. ↑ „Our next-generation model: Gemini 1.5“ (en-us). Google (2024-yil 15-fevral). — „In our research, we've also successfully tested up to 10 million tokens.“.
  116. ↑ Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context (PDF) (Technical report). Google DeepMind. February 15, 2024. Archived (PDF) from the original on Feb 26, 2024. Retrieved May 17, 2024.
  117. ↑ 117,0 117,1 Heikkilä, Melissa; Heaven, Will Douglas (December 6, 2023). „Google DeepMind's new Gemini model looks amazing—but could signal peak AI hype“. MIT Technology Review. December 6, 2023da asl nusxadan arxivlandi. Qaraldi: December 6, 2023.
  118. ↑ Langley, Hugh „Google VP teases Gemini's multimodal future: 'I've seen some pretty amazing things.'“. Business Insider (2023-yil 12-oktyabr). 2023-yil 12-oktyabrda asl nusxadan arxivlangan. Qaraldi: 2023-yil 15-oktyabr.
  119. ↑ Madden, Michael G. „Google's Gemini: is the new AI model really better than ChatGPT?“. The Conversation (2023-yil 15-dekabr). 2023-yil 15-dekabrda asl nusxadan arxivlangan. Qaraldi: 2024-yil 4-fevral.
  120. ↑ Sullivan, Mark (December 6, 2023). „Gemini-powered Google phones may make Siri even more of an Achilles' heel for the iPhone“. Fast Company. December 7, 2023da asl nusxadan arxivlandi. Qaraldi: December 7, 2023.
  121. ↑ Soni, Aditya. „Alphabet soars as Wall Street cheers arrival of AI model Gemini“. Reuters (2023-yil 7-dekabr). 2023-yil 7-dekabrda asl nusxadan arxivlangan. Qaraldi: 2024-yil 4-fevral.
  122. ↑ Swartz, Jon „Gemini, Google's long-awaited answer to ChatGPT, is an overnight hit“. MarketWatch (2023-yil 7-dekabr). 2023-yil 7-dekabrda asl nusxadan arxivlangan. Qaraldi: 2024-yil 4-fevral.
  123. ↑ Kovach, Steve; Elias, Jennifer „Google Faces Controversy over Edited Gemini AI Demo Video“ (en). CNBC (2023-yil 8-dekabr). 2023-yil 9-dekabrda asl nusxadan arxivlangan. Qaraldi: 2023-yil 9-dekabr.
  124. ↑ 124,0 124,1 „Gemini Pro - Technical Details and Benchmarks“. Google DeepMind (2025-yil 25-mart). Qaraldi: 2025-yil 26-mart.

Qoʻshimcha adabiyotlar

[tahrir | manbasini tahrirlash]

Tashqi havolalar

[tahrir | manbasini tahrirlash]
"https://new.ziyosfera.uz/ziyo/Gemini_Deep_Research?oldid=1051784" dan olindi