Google just made its AI lineup simpler and more confusing at the same time. On November 18, 2026, the company launched Gemini 4 Neon, a new model that sits squarely between Hydrogen — the fast, everyday model — and Carbon, the most powerful model in the family, built for the hardest long-horizon work. Neon is the middle child: deeper and more capable than Hydrogen, but faster and cheaper than Carbon. For US users, the question is whether that middle ground is the sweet spot — and whether the $19.99-a-month Pro plan is worth it.
- Gemini 4 Neon launched November 18, 2026, between Hydrogen and Carbon in Google’s element-named AI family.
- API pricing is $3 per million input tokens and $15 per million output tokens, with a 1 million-token context window and no free tier.
- Artificial Analysis benchmarks cited by Google place Neon third overall on the Intelligence Index (61, at xhigh thinking) and first on the Coding Agent Index (58.6).
- Neon is included with Google AI Pro at $19.99 a month, and it replaces Argon in the Gemini app’s model picker.
- Google says Neon is about three times faster than Carbon on some tasks, while matching it on coding, reasoning and agentic work.
What Gemini 4 Neon is
Neon is Google’s answer to a familiar problem: the fastest model is not always smart enough, and the smartest model is not always fast enough. According to its model card, Neon is adapted from Carbon — the flagship — which is why Google says it matches Carbon on coding, reasoning and agentic work, while running about three times faster on some tasks.
That makes Neon the default choice for work that needs real thinking but not the heaviest lifting: deeper analysis, longer documents, multi-step research and coding help beyond a quick answer.
One caveat: Neon’s knowledge cutoff is March 2026, and its full safety report is not out yet — the model card says it “will be available in the coming days.”
Neon vs Hydrogen vs Carbon
Here is how the three models stack up on what Google has actually published:
| Model | In the app | Best for | API price per million tokens | Free tier |
|---|---|---|---|---|
| Hydrogen | Everyday help | Fast, everyday tasks | $1.50 input / $7.50 output | Yes, rate-limited |
| Neon | Deeper analysis | Deeper analysis, coding, agentic work | $3 input / $15 output | No |
| Carbon | Code & complex work | Hardest long-horizon work | Not disclosed in launch materials | — |
Hydrogen, which launched November 12, 2026, is still the speed king of the family — Google says it runs about 2.5 times faster than Gemini 3.8 Flash — and it now powers AI Mode in Google Search, as we covered in Google Gemini 4 in Search: 7 Big Changes US Users Get in Nov 2026. Neon trades some of that speed for much deeper capability, and Carbon trades speed for maximum power.
On benchmarks, Google points to Artificial Analysis, which places Neon third overall on the Intelligence Index with a score of 61 at xhigh thinking, and first on the Coding Agent Index at 58.6. Those are Google’s cited figures, not independent tests run by us — but a number-one coding-agent ranking is a strong signal for developers. The speed claim matters just as much: about three times faster than Carbon on some tasks, according to Google.

Price and who gets it
For developers, Neon costs $3 per million input tokens and $15 per million output tokens through the Gemini API, with a 1 million-token context window. There is no free tier, so developers will need a paid API tier. Neon is also available in Google AI Studio, Gemini for business and Google Antigravity.
For everyone else, access comes through a subscription:
- Google AI Plus — $4.99 a month: Hydrogen only.
- Google AI Pro — $19.99 a month: adds Neon.
- Google AI Ultra — $99.99 a month (5x) or $199.99 a month (20x): adds Carbon.
In the Gemini app, Neon replaces Argon in the model picker for Pro and Ultra subscribers. The picker now labels the three models by job: Hydrogen is “Everyday help,” Neon is “Deeper analysis” and Carbon is “Code & complex work.” Gemini 3.1 Pro is being retired from the app.
Neon can also power Tasks in Gemini, the feature for multi-step jobs that run in the background.
What it means for US users
For most US users, Pro at $19.99 a month is the verdict. It is the cheapest way into Neon, and Neon covers the work most people actually do with AI: long documents, research, analysis and coding help. Paying $99.99 a month for Ultra only makes sense if you regularly need Carbon for the hardest, longest-running work.
The coding angle is the strongest reason to care. A first-place Coding Agent Index ranking, paired with speed close to three times Carbon’s on some tasks, makes Neon a serious option for developers who found Hydrogen too light and Carbon too slow or too expensive.
There is a broader context, too: every major AI company is now racing to ship agentic assistants that do multi-step work on your behalf — a trend we track in our roundup of 7 AI Trends in 2026 — and OpenAI’s Dots assistant is chasing the same idea, as we explain in OpenAI Dots Explained.

The confusing names problem
Here is the part Google has not solved: the names. The Gemini 4 family now uses element names — Hydrogen, Neon, Argon, Carbon — instead of the old Flash and Pro labels users had finally learned. Argon has already been replaced by Neon in the picker, and Gemini 3.1 Pro is being retired from the app, so anyone who memorized the old lineup has to start over.
FAQ
What is Gemini 4 Neon?
How much does Gemini 4 Neon cost?
Is Gemini 4 Neon faster than Carbon?
How do I get Gemini 4 Neon in the Gemini app?
What happened to Gemini 3.1 Pro and Argon?
Gemini 4 Neon is the model most people should use — smarter than Hydrogen, faster and cheaper than Carbon, and bundled into the $19.99 Pro plan. The benchmarks back up the coding claims; the speed claim is the one to watch in real use.
Sources: Google DeepMind Gemini 4 Neon announcement and model card (Nov 18, 2026); Google Gemini API pricing; Artificial Analysis benchmarks as cited by Google.
External sources: Google Blog · Google AI for Developers






