Signal

Gemini 4 Neon costs $3/1M input tokens and $15/1M output tokens; the model sits between Hydrogen and Carbon and replaces Argon in the Gemini app

First reported by 9to5Google ·

The signal ●●●● Compiled by AI from 9to5Google, Techmeme, iPhone in Canada, VentureBeat and Unite.AI
Why you might care

If you use Gemini for complex analysis or coding, you now get faster, higher-quality results at a potentially lower cost than before.

What happened

Google has released Gemini 4 Neon, a new large language model designed to offer Carbon-level quality at Flash speeds and lower cost. It is now available for AI Pro and AI Ultra subscribers within the Gemini app, replacing the Argon model. Gemini 4 Neon sits between the Hydrogen and Carbon models in the Gemini 4 family, excelling in complex coding, analytical reasoning, and enterprise workflows. Performance benchmarks show Neon outperforming comparable models in areas like legal research, tax analysis, and agentic tasks. The model is priced at $3 per 1 million input tokens and $15 per 1 million output tokens. Google is also updating its model picker in the Gemini app, with options including 'Auto', 'Everyday help' (4 Hydrogen), 'Deeper analysis' (4 Neon), and 'Code & complex work' (4 Carbon).

What it means

The introduction of Gemini 4 Neon signals Google's strategy to democratize access to advanced AI capabilities by balancing performance with speed and cost-efficiency. By positioning Neon to deliver Carbon-quality outputs at Flash speeds, Google aims to make sophisticated AI tasks more accessible to a broader range of users and workflows, potentially challenging competitors who offer high-performance models at significantly higher price points or slower speeds.

This move directly impacts users of Gemini's AI Pro and Ultra tiers by offering a more versatile and performant tool for demanding tasks. The replacement of Argon suggests a consolidation and upgrading of the model stack, pushing users towards more powerful options for coding and analysis, while the updated model picker simplifies the user experience by aligning task types with specific model strengths.

AI-written summary. May contain errors.

Crypto