Kivorent News

Wednesday, November 18, 2026

Tech and AINews

Google releases Gemini 4 Neon, a mid-priced model it says nears Carbon at lower cost

Google released Gemini 4 Neon on November 18, a $3-per-million-token model for coding, knowledge work and science that sits between Hydrogen and Carbon. It replaces Gemini 4 Argon in the Gemini app for paying subscribers.

Cortex, covering AI models, labs and agents
, 3 min read

$3per million input tokens; $15 per million output

What to know

  • Google released Gemini 4 Neon on November 18 in the Gemini API, Google AI Studio, Antigravity and the Gemini app for Google AI Pro and Ultra subscribers.
  • The API price is $3 per million input tokens and $15 per million output tokens, twice the price of Gemini 4 Hydrogen, with a 1 million token context window.
  • Google says Neon scored 61 on the Artificial Analysis Intelligence Index and costs 32% less per task than Gemini 4 Carbon at its default setting.
Gemini 4 API list price, per million output tokens
Gemini 4 Hydrogen$7.5
Gemini 4 Neon$15
Gemini 4 Carbon (list)$20

Source: Google, The Decoder and 9to5Google, November 18, 2026. Carbon's list price applies after its introductory discount.

Google released Gemini 4 Neon on Wednesday, November 18, which The Decoder counts as Google's fourth new Gemini model in a month. In a post on Google's blog, Koray Kavukcuoglu, Google DeepMind's senior vice president and Google's chief AI architect, described Neon as the model that "balances depth and efficiency" between the fast Gemini 4 Hydrogen and the top-end Gemini 4 Carbon. Google aims it at software engineering, knowledge work and science tasks.

Price and where it runs

Neon costs $3 per million input tokens and $15 per million output tokens in the Gemini API, according to Google's post. Cached input costs $0.15 per million tokens, a 95% discount. That is twice the price of Hydrogen, which lists at $1.50 and $7.50, and below Carbon's $4 and $20 list price, which takes effect when Carbon's introductory $2 and $10 pricing ends. (For how these rates work, see our explainer on per-million-token pricing.)

The model has a 1 million token context window. It takes text, images, video, audio and PDFs as input and returns text, and it supports computer use, code execution, file search and grounding with Google Search and Google Maps, according to The Decoder and Unite.AI.

Google says Neon is available now in:

  • the paid tier of the Gemini API in Google AI Studio and Android Studio
  • Google Antigravity, for paying Google AI Pro and Ultra subscribers
  • the Gemini app, for AI Pro and Ultra subscribers, from the model menu
  • Gemini for business, through the API and the model menu

In the Gemini app, Neon takes the "Deeper analysis" slot and replaces Gemini 4 Argon, 9to5Google reports, leaving Hydrogen for everyday help and Carbon for code and complex work.

What Google claims

All performance figures below are Google's, and most come from outside leaderboards Google cites rather than tests Kivorent News could check. (Our benchmark explainer covers what such scores do and do not show.)

  • Artificial Analysis Intelligence Index: 61, third overall at the highest thinking setting.
  • Artificial Analysis Coding Agent Index: 58.6, first overall.
  • Vals Index: 70.9%, second overall.
  • Terminal-Bench 4.1: 65.2%. Hydrogen scored 57% on the same test at its launch.
  • Andon Labs Vending-Bench 2: $16,843, first place.

Google says that at the default "high" thinking level Neon costs 32% less per task than Carbon. At its highest setting, Google says it costs at least 36% less per task than competing models at theirs.

Google also says a cyber-defense version, Gemini 4 Neon for cyber, will reach vetted security teams through its Fairwind Program in the coming days. That follows the restricted release of a Gemini 4 Argon model for cyber defenders.

The Gemini 4 lineup

Neon completes a four-model family that Google filled out in quick order, after Argon, Carbon and Hydrogen. Each sits at a different point on price and speed, and Google is now steering most paid work toward Neon, describing it as close to Carbon's quality at lower cost and higher speed. OpenAI released its own two-tier GPT-6.2 Astra and Sol a day earlier.

Why it matters

For developers and companies paying for AI by the token, Neon is the model Google most wants them to use for real work. If its claims about matching Carbon at about a third less per task hold up in outside testing, teams running coding agents and document analysis on Gemini could cut costs without moving to a weaker model. For Gemini app subscribers, it changes which model answers when they ask for deeper analysis.

Sources

GoogleAI modelsAI agents

Written by an AI correspondent under standards set by editor Jacob Knight. Spotted an error? Tell us and we will fix it.