Google unveils flagship Gemini 4 Argon AI model after months of delays
KATHMANDU: Alphabet's Google has unveiled its latest frontier artificial intelligence model, Gemini 4 Argon, ending a 10-month gap since the release of Gemini 3 Pro. The launch follows months of delays attributed to development hurdles concerning the reasoning and coding capabilities of the system.
The technology giant announced the top-tier artificial intelligence (AI) model to anchor its Gemini 4 generation in a fresh bid to catch up to rivals Anthropic and OpenAI in the AI race, Reuters reported. The new model is built to manage complex, long-horizon professional workflows, with particular strengths in software engineering, financial research, and legal drafting.
A central feature of Gemini 4 Argon is its industry-leading context window of one million tokens. This capacity allows the model to process extensive text, code, charts, and lengthy videos in a single trajectory. Furthermore, Argon can autonomously discover, validate, and patch critical software vulnerabilities in a single workflow while minimising human oversight.
Google is taking a cautious approach to the rollout. Initial access is restricted, prioritising vetted experts in cybersecurity defence through the Fairwind Program. The model is currently being tested in controlled enterprise environments for defensive cyber operations before any wider public release.
Reuters said in a report that Google is providing the model to select cybersecurity partners and participating in the United States government's voluntary process for pre-release model access. The news agency noted that the company did not provide a timeline for a broader public release.
Addressing the cybersecurity focus, a Google representative stated, "We really believe that a model of this calibre and this level of frontier performance is meaningfully important for defenders".
When it becomes more widely available to developers and enterprises, introductory pricing for Argon is set at $2 per million input tokens and $10 per million output tokens.
(Compiled)
