Google Launches Gemini 4 Argon, Its Most Advanced AI Model yet - Tech Startups
Google said the model found a vulnerability in hospital software and scored 77.9% on a long-horizon coding benchmark.
- On Wednesday, September 30, Google introduced Gemini Argon, a frontier AI model designed to tackle long-horizon tasks requiring deeper reasoning and enabling enterprise work in legal and finance fields.
- Google significantly expanded Argon's output token limit to one million, up from the previous 64,000 limit, enabling the model to generate hundreds of thousands of tokens in a single trajectory to solve complex problems.
- Argon set a record 77.9 per cent on DeepSWE, tied for first on CWE-bench with 68 per cent, and topped AutomationBench with 51.3 per cent, demonstrating strength across coding and business functions.
- Access remains limited to trusted cybersecurity partners in the Fairwind Programme to prevent misuse, while pricing is set at $2 per million input tokens and $10 per million output tokens with 95 per cent discounts on cached inputs.
- Learning from the standoff Anthropic faced with Mythos—which the Pentagon blacklisted amid concerns regarding potential misuse by intelligence agencies in Russia and China—Google is strictly iterating on safety guardrails before wider rollout.
30 Articles
30 Articles
Google takes a new leap in its artificial intelligence: Gemini 4 Argon arrives with major improvements
Google launches Gemini 4 Argon, its most advanced AI model yet - Tech Startups
Google has unveiled Gemini 4 Argon, its most advanced AI model yet, pushing deeper into coding, cybersecurity, and professional work as the race for frontier AI shifts from chatbots to systems that can carry out long, multi-step jobs. Argon won’t […] The post Google launches Gemini 4 Argon, its most advanced AI model yet first appeared on Tech Startups.
The Gemini 4 Argon generates up to 1 million currents. The model is being gradually launched, starting with the API and Google AI Ultra clients.
Gemini 4, new Google AI, led cybersecurity tests and stood out against prompt injection attacks, in addition to advances in reasoning The post Google launches Gemini 4 and new AI leads digital security tests apareceu primeiro...
Coverage Details
Bias Distribution
- 50% of the sources are Center
Factuality
To view factuality data please Upgrade to Premium
