Technology
Google releases Gemini 4 Argon, restricts access to vetted experts over safety concerns
October 1, 2026 at 05:28 PM · Updated 14d ago
Coverage
4 reports · 3 in Arabic, 1 in English · over 21 hours
يواليوم السابع
كل ما تريد معرفته عن إطلاق Gemini 4 Argon بقدرات ذكاء اصطناعى جديدة
(opens the publisher’s site)
TGThe Guardian
Google rolls out new Gemini AI model but restricts access over safety concerns
(opens the publisher’s site)
يواليوم السابع
جوجل تطلق Gemini 4 Argon بقدرات متقدمة فى البرمجة والأمن السيبرانى
(opens the publisher’s site)SNSky News Arabia
بسبب مخاوف أمنية.. "غوغل" تقيد "جيميناي 4 أرغون"
(opens the publisher’s site)Google has released its new Gemini 4 Argon AI model only to vetted cybersecurity experts and the US government, citing safety concerns.
- Google announced Gemini 4 Argon, withholding the model from the public and releasing it only to cybersecurity experts vetted under its Fairwind program.
- Chief AI architect Koray Kavukcuoglu said safely releasing frontier capabilities at this level requires a phased approach.
- Cybersecurity firm Wiz used the model to uncover a critical hospital software flaw that exposed sensitive personal information.
- Google said it will expand access to paid API clients and Google AI Ultra subscribers after testing, with prices starting at $2 per million input tokens.
- Gemini 4 Argon supports an output limit of up to 1 million tokens per response, Google said.
- Google said the model has been used internally to improve memory in data centers and upgrade large-scale databases.
- The model achieved advanced results on benchmarks including DeepSWE v1.1 and Vals Index, Google said.