
AI-generated summary
Google continues to invest in research and development in the field of AI models. The Gemini series is its flagship large language model family. The Gemini 4 Argon released this time is positioned as a high-performance version that handles complex professional tasks and is designed to compete with other top AI models such as the GPT-6 and Claude series.
Google launches its latest flagship model Gemini 4 Argon. (Picture/official website)
Google has launched a new flagship AI model, Gemini 4 Argon, which focuses on complex tasks such as software engineering, enterprise knowledge work, and network security defense. Enterprise knowledge work also covers legal, financial and other professional fields.
Who gets to use it first?
However, Gemini 4 Argon is currently not available to general users. Google's first wave will only be open to a group of "trusted security defenders" through the Fairwind Program. Next, it will continue to collect feedback from early testers and further strengthen security protection before gradually opening it up to developers, enterprises and general consumers.
Please read on...
Gemini 4 Argon has also been put into use within Google. Google said engineers have used it in their daily work, ranging from general program debugging, large-scale code library migration, to algorithm design.
What is the strength of Gemini 4 Argon?
Google also announced multiple test results of Gemini 4 Argon and competing models such as GPT-6 Astra, Claude Fable 5.1, and Claude Opus 5.5. Judging from the results, Gemini 4 Argon performed well in multiple tests such as knowledge work, financial research, law, long text understanding and multi-modality, while programming had its own winners and losers; in some science and computer operation tests, competitors still had an advantage.
One of the major upgrades of Gemini 4 Argon is its ability to handle extremely long content. Google has increased the upper limit of model output tokens from the original 64,000 tokens to 1 million tokens in one go, allowing AI to have more room to continue reasoning and generating content when faced with long documents, huge codes, or complex tasks that require multi-step processing.
In terms of price, the initial API price of Gemini 4 Argon is US$2 per million input tokens and US$10 for outputs. It will later be adjusted to US$4 for inputs and US$20 for outputs. Caching input tokens will enjoy a 95% discount at the initial stage. Google has not yet announced when the discounted price will end.
"You May Also Want to See"
ChatGPT personal subscription plan major adjustment! The maximum monthly fee for heavy users is 16,500 yuan
AI outlook — possibilities, not facts
Gemini 4 Argon will gradually be available to enterprise customers and developers in the coming months.
Likely · Within months
The API price of Gemini 4 Argon will be adjusted based on market response and competition.
Possible · Within months

A new bug has been reported in Android Auto. Some users of small foldable phones with flip-up and bottom flips cannot make calls through Gemini voice when the phone is closed, and must unfold the phone to return to normal. Google is currently investigating and has not yet announced a fix.

The National Sun Yat-sen Institute of Science participated in the "2026 Taichung AI Application and Smart Manufacturing Exhibition" from October 1 to 4. Focusing on "AI smart drones" and "key precision military technologies", it exhibited 21 achievements including the Jeremiah Series 4, Ruiyuan 2 drones, aerospace power modules and military-standard components, demonstrating its independent research and development strength and cross-domain resilient development.

The legendary SR-71 "Blackbird" high-altitude and high-speed reconnaissance aircraft that has been grounded for nearly 27 years may return to the skies. NASA has moved the 844 airframe at Edwards Air Force Base into the hangar for inspection and contacted former test engineers to evaluate the feasibility of a return flight.

LINE continues to adjust its functions. After stopping services such as VOOM and limited-time updates on October 1, it officially announced that it will officially delete two types of existing posts in Notepad on October 7 to remind users to seize the time to back up important content.

With the support of the "Key Equipment Construction Project of the Institute of Semiconductors" of the National Science Council, National Taiwan University has established an advanced semiconductor process research field on campus to provide one-stop services and open them to other schools to promote industry-university cooperation and international research links. The project has invested more than 400 million yuan to build 71 important equipment, and has cooperated with domestic companies such as TSMC for an amount of approximately 84 million yuan. It has also established a semiconductor research center with the University of Pennsylvania to promote international academic exchanges.

Huawei Technologies is rolling out advanced chips based on its self-developed Tau Scaling Law for the new Mate 90 smartphone series, aiming to surpass 1 million sales and expand into Asia-Pacific, the Middle East, and potentially Europe as part of its effort to overcome US export restrictions through home-grown technology.