TL;DR and context
- Google released Gemini 4 Neon on 18 November at $3 per million input tokens, $0.15 cached and $15 output, with a 1 million token context window.
- Google Cloud lists Neon as generally available under the ID gemini-4-neon, with an EU multi-region endpoint that keeps ML processing in the EU.
- The EU and US endpoints cost 10% more than the global one: $3.30 input and $16.50 output.
- At list price Neon is about 62% cheaper than Claude Fable 5.2 for the same token volumes.
- Google’s cost per task and benchmark claims have not been independently tested.
Gemini 4 Neon starts at $3 input and $15 output per million tokens. Google documents EU ML processing through its supported EU endpoint, where the rates are 10% higher.
Primary document: Google: Introducing Gemini 4 Neon (released 18 November 2026)
Google’s launch post puts Gemini 4 Neon between the faster Hydrogen and the larger Carbon. Gemini 4 Neon pricing is $3 per million input tokens, $0.15 for cached input and $15 per million output tokens, with a 1 million token context window and four thinking levels. The default is high and the top setting is xhigh.
Neon is rolling out in the Gemini API in Google AI Studio and Android Studio for paid developers, in Gemini for business, and in Google Antigravity and the Gemini app for Google AI Pro and Ultra subscribers. A cyber variant goes to approved defenders in Google’s Fairwind Program through their Google Cloud account team.
Where EU requests are processed
The launch post does not mention the EU. Google Cloud’s Gemini 4 Neon model page does. It lists the model as generally available under the ID gemini-4-neon, released 18 November, for verified enterprise customers. It offers a global endpoint and us and eu multi-region endpoints, and states that ML processing on the multi-region endpoints stays in the US or the EU. No single EU country region is listed.
Gemini 4 Neon pricing is higher if data must stay in the EU. Google Cloud’s pricing page lists $3.30 input, $0.165 cached and $16.50 output on non-global endpoints, 10% above global. Batch is half price, at $1.65 and $8.25 outside the global endpoint.
So what
For EU teams, the useful news is not the headline price but the endpoint. A legal or finance team that cannot send client data outside the EU now has a documented route to Gemini 4 Neon through Vertex AI, and it costs a tenth more than the global rate.
IC reading, our arithmetic and not a test we ran: take 10 million uncached input tokens and 1 million output tokens. Gemini 4 Neon pricing gives $45 on the global endpoint and $49.50 on the EU endpoint. At the $8 and $40 rates in our Claude Fable 5.2 pricing brief, the same volumes cost $120. That puts Neon about 62% cheaper on the global endpoint and about 59% cheaper in the EU.
That assumes equal token counts. Thinking tokens are billed as output, so a task run at xhigh can cost far more than one run at high. As our piece on AI token budgets shows, the bill follows tokens per finished task, not the rate card.
Gemini 4 Neon pricing claims Google has not proved
Google says Neon costs 32% less per task than Carbon at the high setting and 50% to 80% less than leading rival models. It cites a score of 61 on the Artificial Analysis Intelligence Index at xhigh, third overall, and first place on Legal Research Bench and Tax Agent Bench. These are Google’s selected results, and we found no independent test of the cost per task claims.
What to watch
- Independent cost tests: whether outside benchmarks confirm Google’s cost per task figures at high and xhigh.
- EU access: how quickly verified enterprise status is granted, and whether single country EU regions follow.
- Retention terms: how long prompts are kept on each endpoint, which the model page does not state.
Sources and limits
This brief rests on Google’s launch post, the Google Cloud model page and the Google Cloud pricing page, all read on 20 November 2026. The cost comparison is Industry Contents’ arithmetic at list prices, not a test. Industry Contents has not run Neon and did not contact Google. Corrections are welcome at corrections@industrycontents.com.
- Google: Introducing Gemini 4 Neon
- Google Cloud: Gemini 4 Neon model page
- Google Cloud: generative AI pricing
- Industry Contents: Claude Fable 5.2 pricing
FAQ
How much does Gemini 4 Neon cost?
$3 per million input tokens, $0.15 cached and $15 output on the Gemini API and the Vertex AI global endpoint. The US and EU endpoints cost 10% more.
Can Gemini 4 Neon process data in the EU?
Yes. Google Cloud lists an eu multi-region endpoint with ML processing in the EU, for verified enterprise customers.
What is the Gemini 4 Neon model ID?
gemini-4-neon, listed as generally available on Google Cloud.
Is Neon cheaper than Claude Fable 5.2?
At list price, about 62% cheaper for the same token volumes. Real costs depend on how many tokens each model uses per task.