Google released its long-delayed Gemini 4 Argon model to paying subscribers this week, ending months of speculation about whether the company could catch up to OpenAI and Anthropic at the frontier of AI development. The model is now rolling out to Google AI Pro and Ultra subscribers in the Gemini app and in AI Mode in Search, as well as to Ultra subscribers on Google's agentic coding platform, Antigravity, according to The Verge. Google has not said when, or whether, Plus subscribers or free users will get access.
Argon is Google's first major frontier model release since Gemini 3 in late 2025, a stretch in which rivals' releases — including OpenAI's GPT-6 Astra and Anthropic's Claude Fable and Opus models — left Google trailing in public perception, compounded by a shakeup in DeepMind's leadership. Google says Argon now matches or beats those rivals across benchmarks spanning coding, cybersecurity, science, finance and law, and that it has already deployed the model internally to find and patch software vulnerabilities, cut the computing cost of moderating YouTube's comments, and debug the flight-physics simulator used by Alphabet's drone-delivery unit, Wing.
A Cybersecurity-First Rollout
Notably, Argon did not go straight to the public. Google first gave access, roughly two weeks before this broader release, to a small group of outside "trusted cyber defenders" through its Fairwind security program, which TechCrunch reported was meant to test the model's ability to "autonomously find, validate, and patch critical software vulnerabilities" before it reached a wider audience. That staged approach — vetting a powerful new model with security specialists before opening it up broadly — has become a standard pattern among frontier AI labs as they try to balance competitive pressure to ship with concerns that more capable models could also hand attackers new tools.
Google is also undercutting rivals on price. Argon's introductory API rate is $2 per million input tokens and $10 per million output tokens — roughly a fifth of what OpenAI and Anthropic charge for comparable top-tier models — though that rate doubles once the introductory period ends in late November. Google also says Argon's maximum output length has expanded far beyond its predecessor, letting it generate much longer single responses for tasks like large codebase migrations.
Still an Open Race
Google frames Argon's arrival as proof the AI race remains unsettled even after a year of OpenAI and Anthropic releases that put Google on the back foot. The timing undercuts any sense of a two-company contest at the frontier: Argon's public debut came just a day after Meta announced its own new frontier model, Muse Wonder Full, underscoring how quickly the competitive order among AI labs continues to shift. Whether Argon actually closes the gap with rivals on real-world coding and reasoning tasks, rather than on the benchmarks Google is citing, will likely become clearer once wider developer access begins in the coming weeks.