OpenAI GPT-6.1 Terra'ni ishga tushirdi — bu kompyuterdan foydalanish, vizual mulohaza, kodlash va kundalik ishga qaratilgan yangi model bo'lib, yirik modellarga nisbatan ancha arzon turadi.
Terra shakllarni to'ldirish, videolarni tahrirlash, texnik chizmalarni talqin qilish va takrorlanuvchi ish stoli ishlarini avtomatlashtirish kabi vazifalar uchun tezroq va arzonroq variant sifatida ishlab chiqilgan.
"Biz GPT-6.1 Terra'ni taqdim etamiz — bu tez, samarali model bo'lib, u aniq vizual mulohaza va aniq kompyuterdan foydalanishni aniqroq, tabiiyroq yozuv bilan birlashtiradi", deb tushuntirdi OpenAI blog postida.
"GPT-5.6 Terra'dan sezilarli darajada yaxshilangan u ish kunini to'ldiradigan aloqa va kompyuterga asoslangan vazifalarda — takliflar yozishdan tortib texnik chizmalarni talqin qilish, shakllarni to'ldirish va videolarni tahrirlashgacha — a'lo natija ko'rsatadi."
OpenAI ma'lumotiga ko'ra, GPT-6.1 Terra API narxi GPT-5.6 Terra va GPT-6.1 Sol'dan 50% pastroq — million kirish tokeni uchun $1 va million chiqish tokeni uchun $5.
Agents' Last Exam'da Terra o'rtacha mulohaza kuchida GPT-5.6 Terra'dan 15 foiz punkt yuqori natija ko'rsatdi, shu bilan birga narxi taxminan uchdan bir qismiga teng.
"U hatto GPT-6 Astra bilan bir xil natija ko'rsatadi, lekin har bir token uchun narxi o'ndan bir qismiga teng", deb ta'kidladi OpenAI. "Terra ancha tezroq ham — u GPT-6 Astra va GPT-6.1 Sol'dan uch baravar tezroq vazifalarni bajaradi."
GPT-6.1 Terra kamroq xatolik (hallutsinatsiya) qiladi
GPT 6.1 Terra kamroq xato qiladi
Ichki benchmarkda past mulohaza kuchida xatolik darajasi GPT-5.6 Terra'dagi 53,6% dan GPT-6.1 Terra'da 16,7% gacha tushdi.
"Maksimal kuchda xatolik darajasi 18 foiz punktga tushib, 4,3% ga yetdi — bu GPT-6 Astra va GPT-6.1 Sol natijalariga deyarli teng", dedi OpenAI.
Kodlash ham yaxshilandi — Terra Terminal-Bench 4.0 da maksimal mulohaza kuchida 50,3% natija ko'rsatdi, bu GPT-5.6 Terra'ning 21,5% natijasidan ikki baravar yuqori.
OpenAI hali ham eng murakkab kodlash vazifalari uchun GPT-6.1 Sol yoki GPT-6 Astra'ni tavsiya qiladi.
GPT-6.1 Terra hozir ChatGPT Work va Codex'da Plus, Pro, Enterprise va Edu foydalanuvchilari uchun, shuningdek API orqali gpt-6.1-terra sifatida mavjud.
Hujumchilar haqiqiy hisob ma'lumotlariga ega bo'lgach, ularning harakatlarining faqat 37% bloklanadi
Umumiy oldini olish ko'rsatkichlari dastlabki kirishdan keyin nima sodir bo'lishini yashirishi mumkin. Hujumchilar haqiqiy hisob ma'lumotlaridan foydalanayotganda, oldini olish keskin pasayadi.
Blue Report 2026 himoya usullarini texnika bo'yicha o'lchaydi — mijozlar ishlab chiqarish muhitlarida o'tkazilgan 338 million simulyatsiya asosida.
Hisobotni oling
OpenAI has launched GPT-6.1 Terra, a new model focused on computer use, visual reasoning, coding, and everyday work, while costing significantly less than its larger models.
Terra is designed as a faster and cheaper option for tasks such as filling out forms, editing videos, interpreting technical drawings, and automating repetitive desktop work.
"We’re introducing GPT-6.1 Terra, a fast, efficient model that combines detailed visual reasoning and precise computer use alongside clearer, more natural writing," OpenAI explained in a blog post.
"A significant step up from GPT-5.6 Terra, it excels at the communication and computer-based tasks that fill the workday, from drafting proposals and interpreting technical drawings to filling out forms and editing videos."
OpenAI says GPT-6.1 Terra's API pricing is 50% lower than GPT-5.6 Terra and GPT-6.1 Sol, at $1 per million input tokens and $5 per million output tokens.
On Agents' Last Exam, Terra scored 15 percentage points higher than GPT-5.6 Terra at medium reasoning effort while costing about one-third as much.
"It even performs on par with GPT-6 Astra at only one-tenth its price per token," OpenAI noted. "Terra is also much quicker, completing tasks over three times faster than both GPT-6 Astra and GPT-6.1 Sol."
GPT-6.1 Terra also hallucinates less
GPT 6.1 Terra makes fewer errors
On its internal benchmark, the error rate at low reasoning effort dropped from 53.6% with GPT-5.6 Terra to 16.7% with GPT-6.1 Terra.
"At maximum effort, the error rate fell by 18 percentage points to 4.3%, nearly matching the rates for GPT-6 Astra and GPT-6.1 Sol," OpenAI said.
Coding has improved as well, with Terra scoring 50.3% on Terminal-Bench 4.0 at maximum reasoning effort, more than double GPT-5.6 Terra's 21.5% score.
OpenAI still recommends GPT-6.1 Sol or GPT-6 Astra for the most complex coding tasks.
GPT-6.1 Terra is available now to Plus, Pro, Enterprise, and Edu users in ChatGPT Work and Codex, as well as through the API as gpt-6.1-terra.
Once attackers have valid credentials, only 37% of their actions are blocked
Overall prevention scores can hide what happens after initial access. Once attackers are using valid credentials, prevention drops sharply.
The Blue Report 2026 measures defenses technique by technique across 338 million simulations run in customer production environments.
Get the report