Gemini 4 Neon: Price, Who Gets It, and Whether It Is Worth Using
Gemini 4 Neon is Google's new middle AI model. Here is what it costs, who can use it, how it compares with Carbon, Hydrogen and GPT-6.2 Sol, and what users say.

Short answer: Gemini 4 Neon is Google's new middle model. It came out on November 18, 2026. If you pay for Google AI Pro or Ultra, you already have it in the Gemini app under "Deeper analysis". For most writing and research jobs it is the one to pick. For simple everyday questions, the cheaper Hydrogen model is enough.
What Neon is
Google now has three main Gemini 4 models, and the names do not explain themselves. One reviewer at Digital Trends called the naming "a complete mess". Here it is in plain words.
| Model | What it is for | Name in the Gemini app |
|---|---|---|
| Hydrogen | Quick everyday tasks | Everyday help |
| Neon | Harder thinking: analysis, research, coding | Deeper analysis |
| Carbon | The hardest, longest projects | Code & complex work |
Neon sits in the middle. Google's pitch is that it thinks almost as well as Carbon but answers faster and costs less.
In the Gemini app, Google says Neon writes more clearly, reads spreadsheets, PDFs and photos more accurately, and can pull details from your Gmail and Docs.
Who gets it
- Gemini app: Google AI Pro and Ultra subscribers. It is rolling out now.
- Developers: the paid Gemini API. It is not on the free API tier.
- Businesses: Gemini for business.
- Free Gemini users: the launch only names paid plans.
Neon also replaced two older options in the app. Gemini 3.1 Pro is gone from the menu, and the model called Argon has been swapped out for Neon.
The price
This only matters if you use the API, which is how apps and automations connect to the model. It charges by the token. A token is a small piece of a word, and 1,000 tokens is roughly 750 words.
| Model | Input, per 1M tokens | Output, per 1M tokens |
|---|---|---|
| Gemini 4 Hydrogen | $1.50 | $7.50 |
| GPT-6.2 Sol (OpenAI) | $2.00 | $10.00 |
| Gemini 4 Neon | $3.00 | $15.00 |
| Gemini 4 Carbon, until November 30 | $2.00 | $10.00 |
| Gemini 4 Carbon, from December 1 | $4.00 | $20.00 |
The odd thing about the price right now
Look at the last two rows. Until November 30, Carbon, the bigger model, is cheaper per token than Neon. Google is running Carbon at a launch price.
On December 1 that ends and Carbon doubles. From then on Neon is 25% cheaper than Carbon.
So if you use the API and want Google's strongest model, the days up to November 30 are the cheap window for Carbon. After that, Neon is the sensible default.
What it costs for real jobs
These are estimates worked out from the prices above.
| Job | Hydrogen | GPT-6.2 Sol | Neon | Carbon from Dec 1 |
|---|---|---|---|---|
| 100 ad scripts | about $0.60 | about $0.80 | about $1.20 | about $1.60 |
| 30 blog posts of 1,500 words | about $0.63 | about $0.84 | about $1.26 | about $1.68 |
| Summary of a 50-page PDF | about $0.06 | about $0.08 | about $0.12 | about $0.16 |
The real bill runs higher, because these models think before they answer and that thinking can be billed as output. But the order stays the same.
The honest reading: for normal creator work, every one of these is cheap. Thirty blog posts for about a dollar. Price should not be what decides it for you. Quality should.
How good is it?
All the scores so far come from Google, so treat them as Google's side of the story until independent tests arrive.
- On the Artificial Analysis Intelligence Index, Google says Neon scores 61. That is ahead of Claude Opus 5.5 at 59 and GPT-6.2 Sol at 57.
- Google says Neon is close to Carbon on the same test while costing 32% less per task.
- It is not ahead everywhere. By Google's own table, Neon is behind Claude Opus 5.5 on two coding tests and behind GPT-6.2 Sol on a chart-reading test.
Thinking levels
Neon has four settings for how hard it thinks: Low, Medium, High and Extra High. The default is High.
Higher levels take longer and cost more on the API, because the model does more thinking before it answers. For a quick caption, Low is fine. For a long research task with many steps, use High or Extra High.
It can take in a lot
Neon can read up to 1 million tokens in one go. That is several long books. You can give it spreadsheets, PDFs, charts, audio and video in the same chat and ask it to work across all of them.
For a creator, that means things like: upload a month of video transcripts and ask which topics got repeated, or give it a full competitor report and your own numbers and ask for the gaps.
What users are saying
The launch thread on Hacker News, a large tech discussion site, has about 80 comments so far. Most people there are developers. Here is where they land.
- The names confuse almost everyone. Several users pointed out there is no clear order to Hydrogen, Neon and Carbon. One (Mogzol) called the names arbitrary.
- Fast and cheaper, but a step below Carbon. One user who tried both (aenis) said Neon is much faster and cheaper than Carbon and clearly weaker in quality. Their advice: use it for carrying out work once the plan is set, not for the planning itself.
- A simple way to split the work. Another (elxr) suggested Hydrogen for simple tasks and Neon as the strong option for harder ones.
- Price is the main complaint. One commenter (maplet) noted Neon scores close to GPT-6.2 Sol on one index but costs more. Another (jrflo) argued OpenAI matches Google at about half the price.
- Doubts about the scores. A few users suspect the AI labs tune their models to do well on the popular tests, so real results may not match.
- Not clear what the $20 plan includes. Commenters disagreed on whether Neon works in Google's coding tool on the cheaper plan.
What to take from it: early users agree with the price table above. Neon is a good middle model, and Sol is the cheaper rival. If you already pay for Google AI Pro, Neon costs you nothing extra to try. If you are choosing an API on price, users are pointing at Sol.
What we do not know yet
- How many Neon messages AI Pro users get per day. Google has not said.
- The maximum length of one answer.
- Whether Neon is coming to AI Mode in Google Search.
- How it does in independent tests.
Who should use which
Use Neon if you:
- Already pay for AI Pro or Ultra and want better answers on research and long documents
- Work across many files at once
- Build with the API and want strong results without Carbon's price after December 1
Use Hydrogen if you:
- Want fast answers to simple questions
- Run high-volume jobs where every cent counts
Look at GPT-6.2 Sol if you:
- Already work in ChatGPT, or want a cheaper API price than Neon
Verdict
If you are on Google AI Pro, switch your hard tasks to "Deeper analysis" and see if the answers improve. It costs you nothing extra. On the API, Neon is fairly priced but not the cheapest, and its best numbers are still Google's own. Worth trying this week, worth judging again when independent tests are out.
