Google introduced Gemini 4 Neon on November 18, 2026. Subscribers to Google AI Pro and Ultra will find the model in the Gemini app. Developers use the model via paid access to the Gemini API in Google AI Studio and Android Studio, and businesses via Gemini for business. In the Antigravity development environment, paying Pro and Ultra subscribers can also select the model.
In Google's model lineup, Neon sits between Carbon for the most demanding projects and Hydrogen for quick everyday tasks. TutKit has already covered both: Carbon and the prices of Hydrogen. According to the plan overview, the Google AI Plus subscription still includes only Hydrogen, and the API's free tier does not include Neon.
Does the launch also apply in Germany?
Google names no exception for Germany or the EU and provides no country list for the Gemini app. According to the region list, the Gemini API and AI Studio are available in Germany, Austria and Switzerland. Google states the API prices only in dollars and gives no euro price.
Where is Gemini 4 Neon included?
Access since November 18, 2026, as Google describes it
Neon is included here
- Gemini app with Google AI Pro or Ultra
- Paid Gemini API in AI Studio and Android Studio
- Gemini for business
- Antigravity with Pro or Ultra
Neon is missing here
- Google AI Plus, which includes Hydrogen
- Free tier of the Gemini API
- Gemini app without Pro or Ultra
Sources: Google Blog, Google AI Studio Docs, Google One
If Neon is missing from the model selection in your app, check your subscription. The announcement lists the Gemini app only for Pro and Ultra, and the Google One plan overview lists the model there.
What can Neon do, and what can't it do?
Neon is meant to help above all with software development, office and research work, and scientific tasks. According to the model page, the model accepts text, images, video, audio and PDF and outputs text. Neon does not generate images or speech. For creatives, the model therefore remains a tool for thinking, analyzing and writing, not an image generator.
On the Artificial Analysis Intelligence Index, Neon at the highest thinking level reaches 3rd place with 61 points according to Google, and 1st place on the Coding Agent Index. In tests on finance, accounting and spreadsheets, Neon is in the lead according to Google. These figures have not been independently verified.
In the API, you pass Word, Excel, PowerPoint and OpenDocument files directly. Instead of the whole document, you specify particular pages, spreadsheets or cell ranges, and according to the release notes, Google bills only the processed part.
What does Neon cost compared to Carbon?
Until November 30, 2026, Neon costs one and a half times as much per token as Carbon and Argon, for which the introductory price still applies. From December 1, according to the pricing page, Google doubles the prices of Carbon and Argon. Neon stays unchanged and is then a quarter below them. Hydrogen costs half as much per token as Neon. For non-urgent jobs, you pay half the standard price in the Batch and Flex tiers, and for priority processing in the Priority tier a surcharge of 80 percent.
At the standard thinking level, Google cites about 32 percent lower costs per task than with Carbon. What your tasks cost only a test will show, because Google bills the tokens for thinking steps as output. According to the documentation, the highest level, xhigh, uses more tokens and takes longer.
What this means for you
If you have Pro or Ultra, open the model selection in the Gemini app and test Neon with a real task, such as analyzing a spreadsheet of sales figures. Still, check the numbers in the result against your source data. The first lesson in the Gemini crash course shows what distinguishes the free version of Gemini from the paid one. With Plus or without a subscription, you don't need to do anything.
If you use the API, according to the developer documentation, the new model ID gemini-4-neon is enough. The default thinking level is then high, whereas for Carbon and Argon it is medium. This increases consumption and waiting time, so set the timeout for responses to at least 15 minutes or let long requests run in the background. Test Neon and Carbon with the same tasks until November 30 and compare the cost per task, not the price per token. Calculate quotes from December onward using the new Carbon prices.
Google unveiled Gemini 4 Neon on November 18, 2026.
Sources