Gemini 4 Argon has a 1M-token output limit, up from 64K for prior models; it initially costs $2/1M input and $10/1M output tokens, rising to $4 and $20 later

Large language model output limits increase to one million tokens, allowing for more detailed and complex generated content.

Google has unveiled Gemini 4 Argon, its latest frontier AI model, featuring a significant increase in output token limit to one million, up from 64,000 in prior models. This enhancement allows for more complex reasoning and problem-solving in a single pass. The model is rolling out in stages, initially to a select group of "trusted cyber defenders" and Google's internal teams without strict guardrails, as part of a phased safety testing approach. Pricing starts at an introductory rate of $2 per million input tokens and $10 per million output tokens, with regular pricing set to increase to $4 and $20 respectively. Cached inputs receive a substantial 95 percent discount. Independent benchmarks suggest Argon performs comparably to OpenAI's GPT-6 Astra and Anthropic's Claude Fable 5.1 on certain reasoning tasks, while exhibiting a notably lower hallucination rate. However, it consumes more tokens than some competitors. Google plans to make Argon available to developers, businesses, and consumers via API access and Google AI Ultra subscriptions as soon as possible after initial testing.

Gemini 4 Argon's one million output token limit represents a substantial leap in generative AI capabilities, potentially enabling new classes of applications that require extensive context and reasoning in a single interaction. This move directly challenges competitors by pushing the boundaries of what AI can achieve in terms of coherent, long-form output, which could redefine user expectations for AI-assisted content creation and analysis.

The tiered pricing structure, with introductory and regular rates, alongside significant discounts for cached inputs, suggests a strategy to balance market adoption with future revenue generation. While initial costs appear competitive, the planned price increase indicates a premium for advanced capabilities, signaling a market trend towards segmenting AI services based on performance and feature sets, influencing how businesses budget for AI integration.