GPT-6 Astra sahifasining manbasini koʻrish

{{Dastur bilgiqutisi
| title = GPT-6 Astra
| logo = 
| screenshot = 
| caption = 
| developer = [[OpenAI]]
| released = {{Start date and age|2026|09|03}} (limited preview)
| replaces = [[GPT-5.6]]
| replaced_by = 
| latest release version = {{Start date and age|2026|09|04}} (stable public release)
| latest release date = 
| programming language = 
| operating system = 
| platform = 
| genre = [[Large language model]]
| license = [[Proprietary software|Proprietary]]
| website = {{URL|https://openai.com/index/gpt-6-astra/}}
}}
{{OpenAI series}}

'''GPT-6 Astra''' — amerikalik [[artificial intelligence|sunʼiy intellekt]] sohasidagi [[OpenAI]] kompaniyasi ishlab chiqqan [[large language model|katta til modeli]] (large language model, LLM). U dastlab 2026-yil 3-sentyabrda tasdiqlangan foydalanuvchilarga taqdim etildi, ertasi kuni esa hamma uchun ochildi.

== Tarixi ==
2026-yil iyul oyidagi [[Hugging Face incident|Hugging Face hodisasi]]dan soʻng OpenAI navbatdagi modelini chiqarishni kechiktirdi. Kompaniya bu vaqt ichida qoʻshimcha himoya choralarini joriy qildi.<ref name=":0" /> GPT-6 Astra 2026-yil 3-sentyabrda, taqdimot kunining oʻzida, cheklangan sinov versiyasi sifatida chiqarildi.<ref>{{Cite web |date=2026-09-04 |title=OpenAI begins rollout of new powerful AI model GPT-6 Astra |url=https://japantoday.com/category/tech/openai-begins-rollout-of-new-powerful-ai-model-gpt-6-astra1 |access-date=2026-09-04 |website=[[Japan Today]] |language=en}}</ref> Kompaniya ertasi kuni modelni pullik foydalanuvchilar uchun ochdi. Bu cheklangan versiya [[cybersecurity|kiberxavfsizlik]] kabi sohalardagi ayrim soʻrovlarni rad etadi.<ref>{{Cite web |last=Brien |first=Jörn |date=2026-09-04 |title=OpenAIs neue KI ist zu gefährlich: Astra kommt abgespeckt |url=https://t3n.de/news/openai-astra-ki-modell-cybersecurity-kritisch-light-version-1761045/ |access-date=2026-09-04 |website=t3n Magazin |language=de}}</ref><ref name=":1">{{Cite web |date=September 3, 2026 |title="Welcome to the AGI era," OpenAI says as GPT-6 Astra debuts |url=https://www.axios.com/2026/09/03/openai-astra-gpt-6-agi-brockman |access-date=2026-09-03 |website=[[Axios (website)|Axios]] |language=en}}</ref>

== Imkoniyatlari ==
OpenAI modelni kiberxavfsizlik, kasbiy faoliyat, dasturiy taʼminot muhandisligi va fan sohalarida «avlodlar oshadigan sakrash» deb baholadi. Kompaniya prezidenti [[Greg Brockman|Greg Brokman]] esa uni kelajakda [[artificial general intelligence|umumiy sunʼiy intellekt]] (artificial general intelligence, AGI) paydo boʻlishi sifatida baholash mumkinligini aytdi.<ref>{{Cite web |last=Field |first=Hayden |date=2026-09-03 |title=OpenAI's next big AI model has 'entered the AGI era' |url=https://www.theverge.com/ai-artificial-intelligence/989601/openai-gpt-6-astra-release |access-date=2026-09-03 |website=The Verge |language=en-US}}</ref><ref>{{Cite web |last=Perlo |first=Jared |date=2026-09-03 |title=OpenAI releases new model that it says triggered internal security measures |url=https://www.nbcnews.com/tech/tech-news/openai-debuts-gpt-6-astra-security-measures-rcna595940 |access-date=2026-09-04 |website=NBC News |language=en}}</ref><ref name=":1" /> OpenAI bir vaqtlar bu tushunchani «iqtisodiy jihatdan qimmatli barcha ishlarni inson darajasida yoki undan yaxshiroq bajara oladigan avtomatlashtirilgan tizim» deb taʼriflagan edi.<ref name=":0">{{Cite web |last=Forlini |first=Emily |title=OpenAI launches GPT-6 Astra, its most powerful model yet, and touts its ability to use your computer |url=https://fortune.com/2026/09/03/openai-debuts-gpt-6-astra-computer-use-greg-brockman-says-start-of-agi/ |access-date=2026-09-03 |website=Fortune |language=en}}</ref> Kompaniyaning taʼkidlashicha, GPT-6 Astra «oldingi har qanday versiyaga qaraganda tezroq va koʻproq vazifani bajara oladi».<ref>{{Cite web |last=Bensinger |first=Greg |date=2026-09-03 |title=OpenAI launches new Astra model amid growing scrutiny over agents' safety |url=https://www.reuters.com/legal/litigation/openai-launches-new-astra-model-amid-growing-scrutiny-over-agents-safety-2026-09-03/ |access-date=2026-09-03 |publisher=[[Reuters]] |language=en}}</ref> Shuningdek, u diqqatni saqlash, vazifa chegaralariga rioya qilish, foydalanuvchi maqsadini tushunish, zerikarli ishlarni bajarish va koʻp bosqichli jarayonlarni yakunlashda ham ustunroq.<ref name=":2">{{Cite web |last=Capoot |first=Ashley |date=2026-09-03 |title=OpenAI begins rolling out Astra model after warning of its advanced cyber capabilities |url=https://www.cnbc.com/2026/09/03/open-ai-astra-gpt-6-cyber.html |access-date=2026-09-03 |website=CNBC |language=en}}</ref> OpenAI GPT-6 Astra bajara oladigan bir necha vazifani sanab oʻtdi: soliq deklaratsiyasini toʻldirish, video oʻyin sahnalarini yaratish, ovqat buyurtma qilish va ish oʻrni izlash.<ref>{{Cite news |last=Booth |first=Robert |date=2026-09-03 |title=OpenAI hails 'new era of artificial general intelligence' with Astra model release |url=https://www.theguardian.com/technology/2026/sep/03/openai-artificial-general-intelligence-astra-release |access-date=2026-09-04 |work=The Guardian |language=en-GB |issn=0261-3077}}</ref> Kompaniya modelni dasturlash, matematika hamda kompyuter va veb-brauzerlarni boshqarishda eng ilgʻor deb taʼrifladi.<ref name=":02" />

OpenAI tadqiqotlar boʻyicha vitse-prezidenti Aydan Klark jurnalistlarga GPT-6 Astra ishlab chiqilishi ular tarixidagi «shubhasiz» eng yirik oʻqitish jarayoni boʻlganini aytdi: «Texasdagi [[Stargate LLC|Stargate]] markazimizda 100 000 dan ortiq [[GPUs|grafik protsessor]]da dastlabki oʻqitish oʻtkazgan birinchi holatimiz».<ref name=":0" />

== Xavfsizlik ==
OpenAI Astrani oʻzining eng [[AI alignment|moslashtirilgan]] modeli deb taʼrifladi,<ref name=":02">{{Cite news |last=Zeff |first=Maxwell |date=2026-09-03 |title=GPT-6 Astra Is Here—and OpenAI Thinks It May Kick Off the AGI Era |url=https://www.wired.com/story/openai-says-gpt-6-can-use-a-computer-better-than-a-human/ |access-date=2026-09-05 |work=Wired |language=en-US |issn=1059-1028}}</ref> ammo uning kiberxavfsizlik imkoniyatlaridan ogohlantirdi.<ref name=":2" /> Kompaniya 2026-yil 1-sentyabrda eng ilgʻor kiberxavfsizlik imkoniyatlaridan foydalanish yanada cheklanishini maʼlum qildi. Ilgʻor kiberxavfsizlik ishlari dastlab sinovchilarning tor doirasiga ochiladi. Keyinchalik Daybreak Blue orqali himoya maqsadidagi foydalanish kengaytiriladi.<ref>{{Cite web |date=September 1, 2026 |title=Path to Astra: critical capabilities and frontier safeguards |url=https://openai.com/index/path-to-astra/ |access-date=2026-09-02 |website=[[OpenAI]] |language=en}}</ref>

GPT-6 Astra «rekurrent chuqurlik» yoki «halqali [[Transformer (deep learning)|transformerlar]]» deb ataladigan yangi [[Reasoning model|mulohaza yuritish]] usulidan foydalanadi. Bu usul samaradorlikni oshiradi, biroq «sunʼiy intellekt mulohazasining, yaʼni „fikrlash zanjiri“ning bir qismini yoki barchasini yashiradigan tarzda ishlaydi».<ref>{{Cite web |last=Kahn |first=Jeremy |title=Why are AI safety experts alarmed by reports OpenAI's Astra model uses "recurrent depth"? |url=https://fortune.com/2026/09/03/reports-openais-astra-model-uses-a-new-more-efficient-ai-architecture-alarms-ai-safety-experts-who-worry-the-method-makes-models-harder-to-control/ |access-date=2026-09-05 |website=Fortune |language=en}}</ref><ref>{{Cite web |date=2026-09-02 |title=OpenAI Technique in 'Astra' Model Sparks Security Concerns |url=https://www.theinformation.com/articles/secret-technique-behind-openais-astra-model-sparks-security-concerns |access-date=2026-09-04 |website=[[The Information (website)|The Information]] |language=en}}</ref> Usul cheklangan tarzda qoʻllansa-da, modelni kuzatib borish imkoniyati borasida xavotir uygʻotdi.<ref>{{Cite web |last=Ulanoff |first=Lance |date=2026-09-03 |title=GPT-6 is here, but what if we just said 'no thanks' to Astra, a model so powerful that we may never fully understand it |url=https://www.techradar.com/ai-platforms-assistants/gpt-6-is-here-but-what-if-we-just-said-no-thanks-to-astra-a-model-so-powerful-that-we-may-never-fully-understand-it |access-date=2026-09-04 |website=[[TechRadar]] |language=en}}</ref><ref>{{Cite web |last=Brandom |first=Russell |date=2026-09-02 |title=OpenAI's new reasoning technique alarms AI safety experts |url=https://techcrunch.com/2026/09/02/openais-new-reasoning-technique-alarms-ai-safety-experts/ |access-date=2026-09-04 |website=[[TechCrunch]] |language=en-US}}</ref> OpenAI bosh olimi [[Jakub Pachocki|Yakub Paxotski]] sunʼiy intellekt keltirishi mumkin boʻlgan koʻzda tutilmagan zararning oldini olish tobora qiyinlashayotganini aytdi. Uning fikricha, bu holat [[AI progress|sunʼiy intellekt taraqqiyoti]]ga toʻsiq boʻlishi mumkin.<ref name=":02" />

== Yana qarang ==
* [[List of large language models|Katta til modellari roʻyxati]]

== Manbalar ==
{{reflist}}

== Havolalar ==
*{{official|https://openai.com/index/gpt-6-astra/}}

{{OpenAI}}

[[Turkum:yilda sunʼiy intellekt]]
[[Turkum:yil dasturiy taʼminoti]]
[[Turkum:Generativ oldindan oʻrgatilgan transformerlar]]
[[Turkum:Katta til modellari]]
[[Turkum:OpenAI]]
[[Turkum:Mulkiy dasturiy taʼminot]]