Power Digital

// AI News

Google releases Gemini 4 Neon: what changed and what it costs

Alex Ng
AI News
Illustration with the headline: Gemini 4 Neon What It Costs

In short: Google released Gemini 4 Neon on 18 November 2026 at US$3 per million input tokens and US$15 per million output tokens (about S$3.87 and S$19.34), with a 1 million token context window. For a typical customer-service agent that is roughly S$137 to S$203 a month in our worked example, depending on whether you use caching.

What happened

Google announced Gemini 4 Neon on 18 November 2026. Google places it in the Gemini 4 family between Carbon, which it describes as the model for the hardest long projects, and Hydrogen, which it describes as the model for fast everyday work. The model ID is gemini-4-neon, and the AI Studio changelog lists it as generally available.

Key details

  • Context window: 1 million tokens.
  • Thinking levels: the Google AI Studio changelog lists low, medium, high (the default) and xhigh.
  • Available in the Gemini API paid tier through Google AI Studio, in Android Studio, in Google Antigravity and in the Gemini app for Pro and Ultra subscribers.
  • Google has not announced Singapore availability.
  • Benchmarks: Google reports a score of 61 on the Artificial Analysis Intelligence Index at xhigh thinking (third place) and first place on the Artificial Analysis Coding Agent Index at 58.6. These are Google's own figures.

Price per million tokens

Converted at US$1 = S$1.2896, derived from the European Central Bank reference rates for 18 November 2026 (EUR 1 = US$1.0968 and S$1.4144). Figures are rounded to the cent.

ModelInput (USD / SGD)Cached input (USD / SGD)Output (USD / SGD)
Gemini 4 Neon$3.00 / S$3.87$0.15 / S$0.19$15.00 / S$19.34
Gemini 4 Hydrogen$1.50 / S$1.93$7.50 / S$9.67
Gemini 4 Carbon (introductory price, valid through 30 November 2026)$2.00 / S$2.58$10.00 / S$12.90

Prices are from the Google AI Studio changelog and the Neon announcement. The Carbon figure is an introductory price. Carbon's price after 30 November has not been announced.

How it compares

Within Google's own line, Neon costs twice as much per token as Hydrogen. Google says Neon uses 32 percent less cost per task than Carbon at high thinking, which matters because cost per task depends on how many tokens a model spends, not only the list price.

What it means for businesses

Worked monthly cost: customer-service agent

Assumptions for this example:

  • 3,000 customer conversations a month.
  • Each conversation sends 10,000 input tokens in total (instructions, product knowledge and the conversation so far, resent each turn).
  • Each conversation produces 1,500 output tokens. We assume any thinking tokens are billed as output, so adjust if your usage differs.
  • That gives 30 million input tokens and 4.5 million output tokens a month.
SetupMonthly cost (USD)Monthly cost (SGD)
Neon, no caching$157.50S$203
Neon, 18M of the 30M input tokens served from cache$106.20S$137
Hydrogen, no caching$78.75S$102
Carbon at introductory price, no caching$105.00S$135

The cached row assumes 6,000 of the 10,000 input tokens per conversation (the fixed instructions and knowledge) are cached. Taxes, hosting, staff review and any free tier limits are not included.

The point for a Singapore business: at this volume the model bill is a few hundred Singapore dollars a month, so the choice between tiers is rarely the largest cost of an agent project. Design, integration and review time usually cost more. If you are scoping a support agent, our AI agent development service starts from this kind of calculation. You can run your own numbers with our token cost calculator.

Our take

This is Power Digital's view, not a test result. For simple, high volume questions, a cheaper tier such as Hydrogen is the sensible place to start, and Neon is worth trying where answer quality on harder cases matters more than the extra cost. Before you move a live agent to a new model, run it on your own conversations and compare quality and cost per resolved case.

Sources

Tags

#ai agents #Google #Gemini 4 #AI pricing

// what we build

Explore Our Services

// AI News

Insights & Resources

Read more →
Back to Articles