AI News
Google Launches Gemini 4 Argon to Take On OpenAI and Anthropic
Google launched Gemini 4 Argon on October 13, 2026, its newest flagship AI model, positioning it as a direct challenger to the latest offerings from OpenAI and Anthropic. The model is built for complex multistep work across software engineering, finance, law, cybersecurity, and science — the high-stakes professional domains where AI labs are now competing hardest.
The launch signals that Google wants Gemini 4 Argon judged on real professional workflows, not just chat benchmarks. Multistep reasoning — planning, tool use, and carrying context across long tasks — is the battleground where the leading labs are differentiating their models for enterprise customers.
API pricing: introductory rates doubling after November 30
API pricing starts at $2 per million input tokens and $10 per million output tokens, with those rates doubling after November 30. The time-limited introductory pricing puts pressure on developers and businesses to evaluate and integrate the model quickly, and it sets up an interesting dynamic for budget planning: early adopters lock in lower costs, while late movers pay twice as much.
For heavy users running agentic workloads that consume large volumes of output tokens, the post-November pricing will matter. The output token rate is the one to watch, since multistep agent work tends to generate far more output than input.
Availability and early testers
Gemini 4 Argon is available to AI Pro and Ultra subscribers in the Gemini app, giving Google's paying users immediate access. Early testers included Deloitte and LexisNexis — two names that tell a story about where Google sees the model's strengths. Deloitte points at consulting and enterprise services; LexisNexis, a legal research giant, points at the legal domain, one of the five professional areas Google is targeting.
With Meta pushing open weights on one side and OpenAI and Anthropic advancing their closed models on the other, Gemini 4 Argon is Google's bid to hold the middle: a flagship closed model aimed squarely at professional and enterprise use. The multistep focus also reflects how enterprise buying is shifting — buyers increasingly pay for agents that complete tasks, not chatbots that answer questions.