Google has introduced Gemini 4 Neon, a new AI model designed for advanced coding, research and enterprise workflows. The model supports a one-million-token context window and starts at $3 per million input tokens.
Google has expanded its Gemini 4 artificial intelligence family with the launch of Gemini 4 Neon, a model designed to handle complex software engineering, analytical work and scientific tasks.
Announced on November 18, 2026, Gemini 4 Neon is positioned between Google’s larger Gemini 4 Carbon model and the lighter Gemini 4 Hydrogen. The company says Neon combines advanced reasoning capabilities with faster task completion and lower operating costs.
One of its most notable features is a one-million-token context window, allowing it to process large collections of documents, spreadsheets, charts, audio and video within a single context.
One million tokens and a new pricing structure

Google Gemini 4 Neon artificial intelligence model promotional graphicGoogle is positioning Gemini 4 Neon as a model for organisations and developers that need to process large amounts of information without relying on the most expensive frontier models.
The published API pricing is:
- $3 per million input tokens
- $0.15 per million cached input tokens
- $15 per million output tokens
The cached input price represents a 95% reduction compared with standard input pricing.
The one-million-token context window is designed to help users analyse lengthy documents, large codebases and multiple information sources in a single workflow.
This could be particularly useful for developers working across several files, researchers analysing extensive material and businesses building automated document-processing systems.
Google claims strong benchmark performance
Google says Gemini 4 Neon achieved a score of 61 on the Artificial Analysis Intelligence Index at its highest thinking level, placing it third overall in the comparison published by the company.
The model also recorded strong results across several specialised benchmarks.
Google reports a 70.9% score on the Vals Index, 84% on RiemannBench and 70.5% on Terminal-Bench Science.
In software development, the company highlights Neon’s ability to debug and refactor code across multiple files, while maintaining performance during long-running, tool-intensive tasks.
These results are based on evaluations cited by Google. Independent testing will be important in determining how the model performs in everyday use and how its capabilities compare across different workloads.
Built for coding, research and enterprise automation
Gemini 4 Neon is designed to support more than traditional question-and-answer interactions.
Google says the model can coordinate complex workflows, work with external tools and transform information from different formats into structured outputs.
For businesses, this could mean analysing financial spreadsheets, reviewing contracts, summarising technical documentation and automating repetitive tasks.
For software developers, its long context window and multi-step reasoning capabilities could help with debugging, codebase analysis and repository-wide changes.
Google also says Neon can support cybersecurity work, including identifying software vulnerabilities, validating risks and helping developers patch weaknesses.
Where can users access Gemini 4 Neon?
Google says Gemini 4 Neon is rolling out through several products and services:
- Developers: Gemini API through Google AI Studio and Android Studio.
- Businesses: Gemini for business and Google Antigravity.
- Consumers: Gemini app access for eligible Google AI Pro and Ultra subscribers.
- Cybersecurity defenders: Access for eligible participants in Google’s Fairwind Program is expected to begin over the following days.
Availability may vary by subscription, product and rollout region.
How does Neon fit into Google’s Gemini 4 family?

Google Gemini 4 Neon artificial intelligence model promotional graphicGoogle describes Neon as a balance between capability, speed and cost.
Gemini 4 Carbon is aimed at the most demanding, long-horizon projects, while Gemini 4 Hydrogen is designed for faster everyday execution.
Neon is intended for complex but frequent workflows where organisations need strong reasoning without paying the cost of using the largest model for every task.
The launch reflects Google’s broader strategy of offering different AI models for different workloads rather than relying on a single system for every use case.
A new challenge for the AI market
The launch of Gemini 4 Neon adds another competitor to an increasingly crowded market for advanced AI models.
The most important questions will be whether its benchmark results translate into practical advantages, how reliably it handles long-running tasks and whether its pricing makes it attractive for businesses deploying AI at scale.
With its million-token context window, coding capabilities and enterprise-oriented pricing, Gemini 4 Neon is designed to appeal to developers and organisations looking for a powerful model that can handle more complex work efficiently.
