Google Launches Gemini 4 Neon: Price, Access and Business Use

· Cortex Solutions

Abstract glowing AI core balancing a light sphere and a dense sphere above a laptop with code, illustrating Google's mid-tier Gemini 4 Neon model

On November 18 Google released Gemini 4 Neon, a mid-tier model for coding, analysis and enterprise work at $3/$15 per million tokens. What it offers and what it means.

On November 18, 2026, Google announced a new model in the Gemini 4 family called Gemini 4 Neon, in an official post on its blog signed by Koray Kavukcuoglu, SVP at Google DeepMind and Chief AI Architect at Google. Tech outlets including 9to5Google and Digital Trends covered the news the same day. Neon arrives roughly a week after the Carbon and Hydrogen models and presents itself as a middle ground between power, speed and cost.

Where does Neon sit in the Gemini 4 family?

According to Google's post, the family has three main tiers:

  • Carbon: for the hardest projects and long, complex tasks.
  • Hydrogen: for fast everyday execution.
  • Neon: balances depth and efficiency, aimed at complex coding, analysis and reasoning, and production-grade enterprise workflows, as well as scientific work.

9to5Google reported that Neon replaced the Argon model in the model picker of the Gemini app, under "Deeper analysis", which Digital Trends also confirmed.

Key specifications

  • A context window of one million tokens, so the model can handle very long documents and codebases in a single request.
  • Four "thinking" levels for developers to choose from: low, medium, high and extra high, with high as the default.
  • According to the Google Cloud developer guide, it supports PDF, plain text and Office files, and offers structured output in a defined format.

Price

In its post, Google set Neon's API price at $3 per million input tokens, $15 per million output tokens and just $0.15 per million cached input tokens, a 95% discount on repeated content. 9to5Google noted that the model is not available on the API's free tier and runs on paid projects only.

Where can it be used?

According to Google, the model is rolling out in:

  • The paid tier of the Gemini API through Google AI Studio and Android Studio.
  • Gemini for business.
  • The Gemini app for AI Pro and Ultra subscribers.
  • Google Antigravity for paid Pro and Ultra subscribers.

The Google Cloud developer guide states that access through the enterprise platform is for "verified enterprise customers in select regions", and Digital Trends said those regions are global, the US and the EU.

Performance and cost according to Google's figures

Google published a long list of benchmark results, including:

  • 61 points on the Artificial Analysis Intelligence Index at the highest thinking level, third overall.
  • Second place on the Vals Index at 70.9%, and first place on Vals' legal research and tax benchmarks.
  • 56.3% on Zapier's AutomationBench, in first place.
  • 58.6 on the Artificial Analysis Coding Agent Index, in first place.

On cost, Google says Neon at the high thinking level costs 32% less per task than Carbon, and 50% to 80% less than leading competing models on some benchmarks. Digital Trends cautioned that these numbers come from Google itself and that independent tests may come out differently, noting that Neon trails some competitors on benchmarks such as Terminal-Bench 4.1 and FrontierCode 2.0.

Safety

Google said Neon is built on the Gemini 4 safety and security stack and carries the same level of protection as Carbon against severe misuse and prompt injection attacks, and it published a model card alongside it.

What does this mean for Saudi businesses?

This is Cortex's reading of the news, not part of Google's announcement:

  • Cheaper capable models: a "mid-tier" model close to top-tier capability at a lower price makes automating everyday work, such as summarising contracts, analysing spreadsheets and answering customer questions, more economically viable for small and medium businesses.
  • Test before you rely on it: the published figures come from the developer, so try the model on your own data and real tasks and compare it with alternatives before building any system on it.
  • Mind data and region: enterprise access is currently limited to specific regions, and companies should review the Personal Data Protection Law and regulators' policies before sending customer data to any external cloud service.
  • Don't tie your systems to one model: models and their prices change every few weeks, so solutions are best built so the model can be swapped without rebuilding the system.

If you are thinking about using AI in your business in a practical and safe way, the Cortex team can help you choose the right model and connect it to your systems: https://cortex.sa/services/ai-automation

Sources: Google's post "Introducing Gemini 4 Neon" (November 18, 2026), 9to5Google (November 18, 2026), Digital Trends (November 18, 2026), and the Google Cloud developer guide for Gemini 4 Neon.

Read next