Google has released Gemini 4 Neon, an AI model it says delivers near top-tier performance on coding, analysis and science tasks at lower cost, priced at $3 per million input tokens, the company said on Wednesday.
The launch adds a cheaper option to Google’s Gemini 4 family as AI developers compete less on raw capability and more on price and speed for businesses that run large volumes of automated work.
What is Gemini 4 Neon?
Gemini 4 Neon sits below Google’s largest model in the family and is designed for long, multi-step tasks such as debugging code across files, analysing spreadsheets and documents, and running “agents” that complete work with little supervision, according to Google’s announcement, written by Koray Kavukcuoglu, chief AI architect at Google.
It can take in up to 1mn tokens, roughly several hundred thousand words, in a single prompt, including text, PDFs, charts, audio and video. Google said Neon offers four “thinking levels”, letting customers trade depth of reasoning against cost.
Gemini 4 Neon is rolling out in the Gemini API through Google AI Studio, to business customers, in the Google Antigravity coding tool for paid AI Pro and Ultra subscribers, and in the Gemini app for Pro and Ultra subscribers, Google said.

How does Gemini 4 Neon compare on price and performance?
Google priced the model at $3 per million input tokens, falling to $0.15 for repeated (cached) inputs, and $15 per million output tokens. Tokens are the small chunks of text that AI models read and write, and the measure by which developers are billed.
The company said Gemini 4 Neon scored 61 on the Artificial Analysis Intelligence Index at its highest setting, ranking third overall behind larger and slower models, and ranked first on the same firm’s coding-agent index. At its default setting, it costs 32% less per task than Gemini 4 Carbon, an earlier model in the family, Google said. These benchmark results were published by Google and have not been independently replicated by DNE Africa.
Google also said Neon recorded its lowest rate of fabricated answers yet in internal testing, and that a version tuned for cyber defence produced more secure code than Carbon. Early business users quoted by Google, including the workplace search company Glean, said the model matched Carbon’s quality at three times the speed.
Why does a cheaper AI model matter?
The cost of running AI models, rather than training them, has become a main concern for companies building products on top of them. Cheaper models with large context windows make it more practical to automate tasks such as document review or customer support at scale, including for start-ups and businesses in markets where budgets are tight.
Gemini 4 Neon arrives in a crowded week for AI releases. Microsoft this week launched cheaper MAI models for coding, security and voice, while concerns about AI tools being turned against computer networks were underlined by an FBI warning about the Artex hacking tool.
FAQ
What is Gemini 4 Neon?
Gemini 4 Neon is an AI model released by Google on 18 November 2026 for coding, analysis and science tasks, positioned as a faster, lower-cost member of the Gemini 4 family.
How much does Gemini 4 Neon cost?
Google priced it at $3 per million input tokens, $0.15 per million cached input tokens and $15 per million output tokens.
Where can I use Gemini 4 Neon?
It is available in the Gemini API in Google AI Studio, for business customers, in Google Antigravity for paid AI Pro and Ultra subscribers, and in the Gemini app for Pro and Ultra subscribers.
How large is its context window?
Google said Neon can process up to 1mn tokens in a single prompt, across text, documents, charts, audio and video.

