Skip to content

AI Tools

What Is Gemini 4 Neon?

Quick Answer

Gemini 4 Neon is Google’s fast frontier model, launched November 18, 2026. It brings frontier-level intelligence with fast end-to-end task completion across complex software engineering, knowledge work and science workflows. It costs $3 per million input tokens and $15 per million output tokens (cached input $0.15, a 95% discount), with a 1-million-token context window and four thinking levels (low / medium / high / xhigh). It is available on the paid tier only — there is no free tier — through the Gemini API in Google AI Studio, Gemini for business, Antigravity (Pro/Ultra) and the Gemini app (Pro/Ultra).

What Is Gemini 4 Neon?

Neon is the middle child of the Gemini 4 family, and Google positions it that way on purpose. Hydrogen (launched November 12) is the fast, low-cost model at $1.50/$7.50. Carbon and Argon are the frontier models for deep, long-horizon reasoning on the hardest problems. Neon sits between them: frontier-level intelligence, but tuned for complex, high-frequency workflows that need to finish fast — repository-level debugging, multi-file refactoring, long agent sessions with many tool calls, and knowledge work on dense documents and spreadsheets.

Google’s own guide places Neon between Hydrogen and Carbon, and the pricing ladder is designed to match: Hydrogen $1.50/$7.50, Neon $3/$15, Carbon $4/$20 from December 1. Until November 30, Carbon’s introductory $2/$10 price is actually cheaper than Neon per token — the ladder only snaps into its intended shape on December 1.

Neon is built on Carbon — Google’s model cards state Carbon is based on Argon and Hydrogen is adapted from Carbon — and it is adapted for speed: same frontier-class capability, faster end-to-end completion.

Release Date and Availability

Google launched Gemini 4 Neon on November 18, 2026, in a blog post by Koray Kavukcuoglu, SVP at Google DeepMind and Chief AI Architect at Google. It went live across most developer and consumer products the same day.

It is rolling out now through:

  • Gemini API in Google AI Studio — paid tier, model id gemini-4-neon
  • Gemini for business — via API and model dropdown
  • Google Antigravity — paid Google AI Pro and Ultra subscribers
  • Gemini app — Pro and Ultra subscribers
  • Android Studio — via the Gemini API

A separate cyber variant, Gemini 4 Neon for cyber, is rolling out to trusted defenders in the Fairwind Program in the coming days. The launch post makes no mention of Vertex AI.

Key Gemini 4 Neon Features

1. Frontier intelligence, fast completion

Google pitches Neon as frontier-level intelligence with fast end-to-end task completion — not the deepest reasoning in the family, but the fastest way to finish complex work.

2. Four thinking levels

Low, medium, high (default) and xhigh set how deeply the model thinks — and how many tokens it spends. Google suggests low for fast, high-volume work, medium for interactive coding, high as the default, and xhigh only for hard, multi-step problems where high falls short, since it uses more tokens and takes longer.

3. A 1-million-token context window

1,048,576 tokens, shared by input, thinking and output. Enterprise work — spreadsheets, PDFs, charts, audio and video — can be synthesised across all of those inputs in a single context window and turned into structured deliverables.

4. Long decode

Maximum output is 65,536 tokens by default, and up to 1,048,576 tokens with long decode — the model can write very long outputs when a task needs them.

5. No free tier, batch at half price

Neon is not available on the free tier at all — every token is billed, thinking tokens included. The Batch API is exactly half the standard price ($1.50/$7.50), and Priority is $5.40/$27. Caching is implicit only; explicit caching is not supported.

6. Ranked #3 overall on Artificial Analysis

At its xhigh thinking level, Neon scores 61 on the Artificial Analysis Intelligence Index, ranking #3 overall — just behind larger, slower models. That is Google’s cited third-party ranking, not its own benchmark.

Gemini 4 Neon Pricing Explained

Standard paid-tier rates, per million tokens, from Google’s Gemini API pricing page.

Item Neon price
Input tokens $3.00 per 1M tokens
Output tokens $15.00 per 1M tokens
Cached input tokens $0.15 per 1M tokens (95% discount)
Batch API $1.50 / $7.50 per 1M tokens
Priority tier $5.40 / $27.00 per 1M tokens
Free tier Not available

A worked example: a task using 1 million input tokens and 500,000 output tokens costs $3 + (0.5 × $15) = $10.50 at standard price, or $5.25 on the Batch API. The same task on Hydrogen would cost $5.25 standard — half of Neon — which is why Hydrogen stays the pick for fast everyday execution.

The pricing puzzle is Carbon. Through November 30, Carbon costs $2/$10 — cheaper per token than Neon. From December 1, Carbon doubles to $4/$20, and Neon becomes 25% cheaper per token than Carbon. Google’s answer is that cost per task matters more than cost per token: at its default high thinking level, Neon costs 32% less per task than Carbon, it says, even while Carbon is cheaper per token this month.

Gemini 4 Neon vs the Gemini 4 Family

Spec Hydrogen Neon Carbon (from 1 Dec)
Launched Nov 12, 2026 Nov 18, 2026 Earlier
Input price per 1M $1.50 $3.00 $4.00
Output price per 1M $7.50 $15.00 $20.00
Context window 1M tokens 1M tokens 1M tokens
Thinking levels 4 4 (low/medium/high/xhigh) 4
Free tier Yes No No
Best for Fast everyday execution Complex, high-frequency workflows Deepest long-horizon reasoning

Neon costs exactly twice Hydrogen per token, and from December 1 it costs three-quarters of Carbon. Until November 30 it costs 50% more per token than Carbon’s introductory price — so anyone comparing the three this month should check the date, not just the price.

Performance and Benchmarks

All performance figures here are Google’s own claims from its launch post, unless marked otherwise, and have not been independently verified by faqsnow.

Google says Neon ranks ahead of comparable models and #3 overall with a score of 61 on the Artificial Analysis Intelligence Index at its xhigh thinking level, landing just behind larger, slower models. At its default high thinking level, Google says Neon costs 32% less per task than Carbon. Google also claims Neon costs roughly 50% to 80% less per task than leading rival frontier models on the Vals Index and Agent Arena.

The model launched on November 18, so no independent reproduction of those cost-per-task claims exists yet. Treat the table as the manufacturer’s spec sheet, not a review — the ranking (Artificial Analysis, third-party) is the strongest independent signal so far.

How to Use Gemini 4 Neon

1. Use the Gemini API in Google AI Studio (paid tier)

Open AI Studio, switch to the paid tier, and select gemini-4-neon. There is no free tier, so even a test call is billed — start with a small input.

2. Pick the thinking level by the job

Use low for fast, high-volume work, medium for interactive coding, high (the default) for most complex work, and xhigh only for hard, multi-step problems where high falls short. Higher levels use more tokens and take longer, so they cost more per task.

3. Use Batch for slow jobs

If a job can wait, the Batch API costs half the standard price — $1.50/$7.50 — which makes Neon the same price as Hydrogen’s standard rate.

4. Use it in Antigravity or the Gemini app if you are a Pro/Ultra subscriber

Paid Google AI Pro and Ultra subscribers get Neon in Antigravity and in the Gemini app, and in Gemini for business via the API and model dropdown.

5. Mind the Carbon price change on December 1

If you are running Carbon at its $2/$10 introductory price, it doubles to $4/$20 on December 1 — at that point Neon becomes the cheaper option per token as well as per task (on Google’s claim). Plan the switch before then, not after.

Best Uses

  • Complex coding — repository-level debugging and multi-file refactoring
  • Long agent sessions — many tool calls in a single session, inside the 1M-token window
  • Knowledge work — dense documents, spreadsheets, PDFs, charts, audio and video synthesised into structured deliverables
  • Science workflows — analytical reasoning across long sources
  • High-frequency complex work — where Carbon is too slow or too expensive per task, and Hydrogen is not strong enough

Limitations and Things to Know

  • No free tier. Every token is billed, thinking tokens included — the cheapest way to try it is the $1.50/$7.50 batch tier.
  • Carbon is cheaper per token until November 30. Carbon’s $2/$10 introductory price beats Neon’s $3/$15 until it doubles on December 1. Neon’s cost advantage is per task (Google’s claim), not per token, this month.
  • Output pricing includes thinking tokens. There is no separate thinking charge, which matters for a reasoning model — a high or xhigh run bills more output tokens.
  • Explicit caching is not supported. Only implicit caching, at $0.15 per million tokens.
  • Benchmarks are Google’s own. The 32%-less-per-task and 50–80%-less claims come from Google; the Artificial Analysis #3 ranking is third-party.
  • Vertex AI was not mentioned. The launch post does not say Neon is on Vertex AI; enterprise teams on Google Cloud should check before assuming it is there.
  • The cyber variant is separate. Gemini 4 Neon for cyber rolls out only to trusted defenders in the Fairwind Program, not generally.

Frequently Asked Questions

What is Gemini 4 Neon?

Gemini 4 Neon is Google’s fast frontier model, launched November 18, 2026. It brings frontier-level intelligence with fast end-to-end task completion for complex software engineering, knowledge work and science workflows, at $3/$15 per million tokens.

How much does Gemini 4 Neon cost?

$3 per million input tokens and $15 per million output tokens on the paid tier, with cached input at $0.15 (95% discount). Batch is half price ($1.50/$7.50); Priority is $5.40/$27. There is no free tier.

How is Neon different from Hydrogen and Carbon?

Hydrogen ($1.50/$7.50) is the fast, low-cost model for everyday execution. Carbon is the frontier model for the deepest long-horizon reasoning — $2/$10 through November 30, then $4/$20 from December 1. Neon ($3/$15) sits between them: stronger than Hydrogen, faster and cheaper per task than Carbon (on Google’s claim).

What is the context window?

1,048,576 tokens, shared by input, thinking and output. Maximum output is 65,536 tokens by default, up to 1,048,576 with long decode.

What are the thinking levels?

Four: low, medium, high (default) and xhigh. Low is for fast, high-volume work; medium for interactive coding; high for most complex work; xhigh only for hard, multi-step problems.

Where can I use Gemini 4 Neon?

Through the Gemini API in Google AI Studio (paid tier), Gemini for business, Google Antigravity (Pro/Ultra), the Gemini app (Pro/Ultra), and Android Studio. The cyber variant is Fairwind Program only.

Is Gemini 4 Neon free?

No. It is not available on the free tier at all. The cheapest way to try it is the Batch API at $1.50/$7.50 per million tokens.

What is the model id?

gemini-4-neon.

Final Verdict

Neon is the most practical Gemini 4 model for people who actually ship work: frontier-class intelligence without Carbon’s price or slowness, and a real step up from Hydrogen when the work gets complex. The pricing needs a calendar, not just a calculator — Carbon is cheaper per token until November 30, then Neon is cheaper from December 1 — so the smart move is to try Neon on Batch now and plan the Carbon switch before December. The honest limit is that the cost-per-task claims are Google’s own and unreproduced a day after launch; the Artificial Analysis #3 ranking is the best independent evidence so far, and it is genuinely strong for a model pitched on speed.

Editorial update: November 19, 2026. This article is based on Google’s official launch post “Introducing Gemini 4 Neon” (18 Nov 2026) and Google’s Gemini API pricing page. Performance figures are Google’s own unless marked third-party, and have not been independently verified by faqsnow.