Short answer: Gemini 4 Neon is Google's fast frontier model in the Gemini 4 family, released on November 18, 2026. It sits between Gemini 4 Hydrogen, the everyday speed option, and Gemini 4 Carbon, the model Google saves for the hardest long projects. On the paid Gemini API it costs $3 per million input tokens, $0.15 per million cached input tokens and $15 per million output tokens, with a 1 million token context window, checked on November 20, 2026 (Google, Gemini API pricing). It is not on the free API tier.
If you just saw "Neon" or gemini-4-neon in a model menu, this guide covers what it's for, what it costs, who can use it, how it works as an Orchestrator in Google Antigravity, and when Hydrogen, Argon or Carbon is still the better pick.
What is Gemini 4 Neon?
Google describes Neon as a frontier level model that finishes complex software, knowledge work and science tasks quickly. The DeepMind model card says Carbon handles your hardest, long horizon projects, Hydrogen handles fast everyday execution, and Neon balances depth and efficiency (Gemini 4 Neon model card).
A few specs from Google's developer docs, checked November 20, 2026 (Gemini 4 Neon, API changelog):
- Model ID:
gemini-4-neon - Context window: one shared window of 1,048,576 tokens for input and output
- Output: up to 65,536 tokens by default, and up to the full window with long decode
- Inputs: text, images, audio, video and PDFs
- Thinking levels:
low,medium,high(the default) andxhigh
Google also says Neon can act as a fast Orchestrator in Antigravity, planning the work and directing Hydrogen subagents in parallel.
Where can you use it, and is it free?
No, Neon is a paid model. Google's launch post and the Gemini API pricing page both say it is not available on the free API tier. Here's where Google says it began rolling out on November 18, 2026:
- Gemini API and Google AI Studio: paid tier only. The model code is
gemini-4-neon. - Google Antigravity: paid Google AI Pro and Ultra subscribers, in the app, CLI and IDE extensions.
- Gemini app: Pro and Ultra subscribers can pick it from the model menu.
- Gemini for business: through the API or the model dropdown.
How much does Gemini 4 Neon cost?
On the paid Gemini API, Google's pricing page lists a flat rate across the 1 million token window, checked November 20, 2026:
| Token type | Price per million tokens |
|---|---|
| Input | $3.00 |
| Cached input | $0.15 (implicit caching only; explicit caching is not supported) |
| Output, including thinking tokens | $15.00 |
Thinking tokens count as output, so xhigh costs more than high for the same prompt.
List price alone can mislead you when you compare families. Argon and Carbon still have an introductory API price of $2 per million input tokens and $10 per million output tokens through November 30, 2026. On December 1, 2026 those rates become $4 and $20. Until that date, Neon's list price is actually higher than the Argon and Carbon promo. After it, Neon is cheaper per token than those two. Google separately says that at Neon's default high thinking level, cost per task is 32% lower than Carbon. That's a vendor figure, not the list price.
Hydrogen remains the inexpensive option: $1.50 in and $7.50 out on the paid tier, and it is also on the free tier.
What are Neon's thinking levels?
The thinking level is how much the model reasons before it answers. Neon has the same four levels as Hydrogen:
| Level | What Google says it's for |
|---|---|
low | Quicker, cheaper turns |
medium | A step up from low |
high (default) | Most daily work |
xhigh | The hardest engineering, analysis and science jobs |
Google's launch post says Neon posts its highest intelligence score at xhigh. Start with high unless the task clearly needs more.
How does Orchestrator work with Neon?
In Google Antigravity, Orchestrator pairs two Gemini 4 models in one conversation. Neon can be the planner: it reads the repo, writes the plan and reviews the result, while Hydrogen does the implementation, such as code edits, package updates and tests. If Hydrogen gets stuck, it hands control back to Neon (Antigravity blog).
Google says this is for people who want Neon's planning with Hydrogen's speed, without switching models by hand. Orchestrator (Neon) is rolling out to the same paid Pro and Ultra Antigravity plans that get Neon itself.
When should you pick Neon, Hydrogen, Argon or Carbon?
A simple way to choose, based on Google's own roles and the list prices checked on November 20, 2026:
- Pick Hydrogen for free access, everyday chores and a lower API bill. It's the fast default.
- Pick Neon for frontier quality coding, analysis and science work that you still want to finish quickly. It's paid only.
- Pick Argon for long, advanced tasks, including security work, when you want fewer tokens than Carbon.
- Pick Carbon for the hardest, longest projects.
If a vendor score would change your mind, Google reports Neon at 61 on the Artificial Analysis Intelligence Index at xhigh, at #1 on the Artificial Analysis Coding Agent Index (58.6), and at #2 on the Vals Index (70.9%). Those are Google's cited figures. Try the same real task on Hydrogen and Neon before you change a default.
Go deeper
- What is Gemini 4 Hydrogen: the fast Gemini 4, where it's free, and how Orchestrator works.
- Gemini 4 Argon: the larger model, including how it handles security work.
- AI tools for developers: the 2026 tools, sorted by what they do.
The skill behind picking a model
New Gemini 4 names now arrive a few days apart. The useful skill is knowing which one fits the job, what it costs on your plan, and how to test it on real work. If you want to get better at building with models and agents, at your own pace, see how AI Flex works.
Sources, checked November 20, 2026: Introducing Gemini 4 Neon (Google, November 18, 2026), the Gemini 4 Neon model card, Gemini 4 Neon in Google Antigravity, Gemini API pricing, the Gemini 4 Neon API docs and the Gemini API changelog.
