Gemini 4 Neon: Google’s New AI Model, Price, Features and How to Try It
Google has launched Gemini 4 Neon, a model between Hydrogen and Carbon priced at $3 and $15 per million tokens. What it does, how it compares and how to try it.
Image: TeQBlogs
Gemini 4 Neon is Google’s newest AI model, announced on 18 November 2026. It sits between the fast Gemini 4 Hydrogen and the heavyweight Gemini 4 Carbon, and Google pitches it as frontier-level intelligence at a lower cost per task. Developers can use it through the paid Gemini API at $3 per million input tokens and $15 per million output tokens, and Google AI Pro and Ultra subscribers can pick it in the Gemini app.
It is the fourth Gemini 4 model in about seven weeks, so the obvious question is which one you should actually use. The short answer: Neon is meant to be the default for serious work, the model you reach for when Hydrogen is not quite clever enough but Carbon is slower and dearer than the job needs.
Quick summary
| Detail | Gemini 4 Neon |
|---|---|
| Announced | 18 November 2026 |
| Position | Between Gemini 4 Hydrogen (fast) and Gemini 4 Carbon (hardest tasks) |
| API price | $3 per million input tokens, $15 per million output tokens |
| Cached input | $0.15 per million tokens (95% discount) |
| Context window | 1 million tokens |
| Thinking levels | Four |
| Gemini app | Google AI Pro and Ultra subscribers, via the model menu |
| Developers | Gemini API paid tier, Google AI Studio, Android Studio, Google Antigravity |

Where Gemini 4 Neon fits in the Gemini 4 family
Google now has three tiers that matter for most people. Hydrogen is for fast everyday tasks. Carbon handles the hardest long-running projects. Neon sits in the middle and, according to Google, gives up very little intelligence while costing noticeably less per task.
Also read: HDD vs. SSD: Which one should you go for?
| Model | Best for | Trade-off |
|---|---|---|
| Gemini 4 Hydrogen | Quick answers, everyday tasks, high volume | Less depth on complex problems |
| Gemini 4 Neon | Coding, analysis, research and multi-step work | Paid plans only |
| Gemini 4 Carbon | The hardest long-horizon projects | Slower and more expensive per task |
Google says Neon at its default “high” thinking level costs 32% less per task than Carbon. Developers can also have Neon direct several Hydrogen subagents working in parallel in Google Antigravity, which is how Google expects big coding jobs to be split up.
What Google says it can do
These figures come from Google’s own announcement. Independent testing will take a few weeks, so treat rankings as Google’s claims for now.
- Coding: first on the Artificial Analysis Coding Agent Index with 58.6, and 65.2% on Terminal-Bench 4.1.
- Knowledge work: first on the Legal Research and Tax Agent benchmarks in the Vals Index, and first overall on Zapier AutomationBench.
- Science: 70.5% on Terminal-Bench Science and 84.0% on RiemannBench, which Google calls best in class.
- Accuracy: the lowest fabrication rate Google has recorded in its internal testing.
- Security work: state-of-the-art scores on several code security benchmarks, with stronger secure-coding results than Carbon.
Is it safe?
Google says Neon is built on the Gemini 4 safety stack with the same protections as Carbon against severe misuse and prompt injection, but refuses far less legitimate work. It reports a 0.09% misaligned-action rate on its agentic safety tests and a 99.2% pass rate on an honesty evaluation. These are Google’s own numbers; the full detail is in the model card.
How to try Gemini 4 Neon

- Gemini app: if you subscribe to Google AI Pro or Ultra, open Gemini and choose Neon from the model menu.
- Google AI Studio: Pro and Ultra subscribers can try it in AI Studio without writing code.
- Developers: call it through the Gemini API on the paid tier, or use it in Android Studio and Google Antigravity.
- Businesses: it is available in Gemini for business through Google Cloud’s Gemini Enterprise.
Google’s announcement does not list country-by-country availability. If Neon does not appear in your model menu in India, check that your plan is Google AI Pro or Ultra and update the app.
What it means for Indian users and businesses
- Cheaper serious AI: the 95% discount on cached input makes long documents and codebases much cheaper to work with repeatedly, which matters for small teams on tight budgets.
- Tasks get faster: Google says Tasks in Gemini, which is free in India, can run on Neon for faster multi-step jobs.
- The price war continues: Neon arrives days after GPT-6.2 Astra and alongside rivals like Qwen4-Max. Compare on your own tasks before switching tools.
If you are rolling AI agents out at work, read our guide to AI agent security first. See also Gemini 4 Hydrogen and Gemini 4 Carbon.
Frequently asked questions
What is Gemini 4 Neon?
It is a Google AI model announced on 18 November 2026 that sits between Gemini 4 Hydrogen and Gemini 4 Carbon, aimed at coding, knowledge work and science at a lower cost per task.
How much does Gemini 4 Neon cost?
Through the Gemini API it costs $3 per million input tokens, $0.15 per million cached input tokens and $15 per million output tokens.
Is Gemini 4 Neon free?
No. In the Gemini app it is available to Google AI Pro and Ultra subscribers, and in the API it is on the paid tier.
What is the context window of Gemini 4 Neon?
Google lists a context window of 1 million tokens.
The bottom line
Gemini 4 Neon is Google’s answer to the question most teams actually ask: which model is good enough for real work without the top-tier price? On Google’s own numbers it comes close to Carbon at a lower cost. Try it on your own tasks before you rewrite your workflow around it, because benchmark wins do not always survive contact with real projects.
