Google has released Gemini 4 Argon to developers, businesses and paid subscribers worldwide after an initial preview for selected cybersecurity partners. The launch puts Google’s newest flagship model into coding, research, enterprise and consumer-facing tools as competition with OpenAI and Anthropic intensifies.
Global access begins
According to Reuters, Argon is available through Google’s AI development tools, Google Cloud and the Gemini API. Google AI Pro and Ultra subscribers can access it in the Gemini app and AI Mode in Search, while the Antigravity coding platform is limited to Ultra subscribers.
Google says Argon is designed for long, multi-step work in software development, finance, legal research, cybersecurity and science. The company’s launch announcement says the model ranks first on the Vals Index, a claim based on Google’s cited evaluations.
Pricing doubles after November
The model launches at $2 per million input tokens and $10 per million output tokens, with cached input tokens discounted by 95 percent. Google says those introductory rates end on November 30, 2026, when pricing will rise to $4 per million input tokens and $20 per million output tokens.
Google describes Argon as having a unified one-million-token input-and-output limit. The company says the model is available under the identifier gemini-4-argon and can process long documents containing text, tables, charts and images.
Enterprise work gets longer
Google is positioning Argon as an agentic enterprise model for extended, multi-step tasks. The company says internal teams using it through Antigravity increased agentic code submissions by more than 35 percent and reduced code rollbacks by nearly one-third. Google Cloud also reports that Argon-powered changes to its Agent Development Kit reduced long-session preparation time by 40 percent and improved cold-import speed by 21 percent.
Those figures are company-reported. Argon is being distributed through consumer subscriptions, cloud services and developer APIs, giving organizations several routes to evaluate the model.
Cybersecurity access expands
Google is expanding its Fairwind Program to hundreds of additional vetted cybersecurity partners. The program provides selected defenders with access to Argon without cyber guardrails, while standard deployments retain restrictions on offensive cybersecurity tasks.
Google says Argon reduced attack success to 0.7 percent on a 15-attempt indirect prompt-injection benchmark. Its Google Cloud announcement also says the model tied for first on CWE-Bench v1 for vulnerability remediation.
Robotics reaches 87 percent
Google’s developer team reports an 87 percent score for Argon on the Embodied Reasoning benchmark, which tests physical and spatial problem-solving. In a separate demonstration, Google said the model watched a video of a person playing cornhole, built a simulation and transferred the learned skill to Apptronik’s Apollo humanoid robot. The robotics report describes the result as an example of converting ordinary video into robot-training simulations.
Google is also offering 10,000 scientists six months of free access to Antigravity’s highest usage tier, subject to eligibility and capacity.
References
Google. (2026, October 13). Gemini 4 Argon: Built to solve your hardest problems. https://blog.google/intl/en-mena/product-updates/explore-get-answers/gemini-4-argon-built-to-solve-your-hardest-problems/
Gerstenhaber, M. (2026, October 13). Gemini 4 Argon available on Google Cloud, built to help solve your business’ hardest problems. Google Cloud Blog. https://cloud.google.com/blog/products/ai-machine-learning/gemini-4-argon-available-on-google-cloud
Rao, K. (2026, October 13). Gemini 4 Argon for robotics. Google Developers Blog. https://developers.googleblog.com/en/gemini-4-argon-for-robotics/
Reuters. (2026, October 13). Google launches flagship AI model Gemini 4 Argon for developers, businesses. Investing.com. https://www.investing.com/news/stock-market-news/google-launches-flagship-ai-model-gemini-4-argon-for-developers-businesses-4947240