On this page
What is Gemini 4 Neon?
Gemini 4 Neon is a mid-tier model in Google's Gemini 4 family, announced on 18 November 2026. Google says it delivers frontier-level reasoning for coding, knowledge work and science at lower cost than Gemini 4 Carbon. It is open to AI Pro and AI Ultra subscribers, and API pricing is $3 per million input tokens and $15 per million output.
Google published the announcement on its blog under the name of Koray Kavukcuoglu, its chief AI architect, and posted a model card on the DeepMind site the same day.
Where does Gemini 4 Neon sit in the Gemini 4 line-up?
Google positions Neon between two models we have already covered. Carbon handles the hardest long-horizon projects, Hydrogen handles fast everyday execution, and Neon, in Google's words, balances depth and efficiency. For earlier coverage, see our reports on Gemini 4 Carbon opening to paying users and Gemini 4 Hydrogen in Google Search.
The model card says Neon is adapted from Gemini 4 Carbon. It has a knowledge cutoff of March 2026 and takes text, images, audio and video files as input.
9to5Google reports that the Gemini app model picker now reads Auto, Everyday help (Hydrogen), Deeper analysis (Neon) and Code and complex work (Carbon). It also reports that Neon replaces Argon in the app and that Gemini 3.1 Pro is being removed. Google's own blog post does not say Neon replaces anything, so treat the Argon point as 9to5Google's reporting.
What can Gemini 4 Neon do, and what does it cost?
Google lists the following core specifications on its blog and model card:
| Detail | Gemini 4 Neon |
|---|---|
| Context window | 1 million tokens |
| Input price | $3 per million tokens |
| Cached input price | $0.15 per million tokens |
| Output price | $15 per million tokens |
| Thinking levels | Four, with high as the default and xhigh above it |
| Default synchronous output | Up to 64,000 tokens |
Google says that at its default high setting Neon costs 32% less per task than Carbon. That is a company figure, not an independent measurement. 9to5Google adds that Neon is twice the price of Hydrogen, and that Argon and Carbon are priced at $2 and $10 per million tokens until 30 November before rising to $4 and $20. On those figures Neon would be about 25% cheaper than Carbon once the promotion ends.
Digit also reports a Google claim that Neon is three times faster than Carbon, though it gives no independent speed tests.
How does Gemini 4 Neon score on benchmarks?
All of the numbers below come from Google's announcement and model card. Some cite third-party indices, but Google chose which results to show.
- Artificial Analysis Intelligence Index: 61 at xhigh, which Google says is third overall, just behind larger and slower models.
- Artificial Analysis Coding Agent Index: first overall, at 58.6.
- Science and maths: best in class on Terminal-Bench Science (70.5%), RiemannBench (84.0%) and HLE-Diamond (71.3%), according to Google.
- Agents: second on the Agent Arena Index with a score of 1336 across more than 4,200 sessions, and first on Zapier's AutomationBench at 56.3%.
The model card compares Neon with Anthropic's Claude Opus 5.5 and OpenAI's GPT-6.2 Sol. Neon leads on most rows. Claude Opus 5.5 is slightly ahead on Terminal-bench 4.1 (65.8% against 65.2%) and FrontierCode 2.0 (47.3% against 46.2%), and GPT-6.2 Sol is ahead on the MKA Chartography chart-reading test (75.1% against 73.5%).
Who can use Gemini 4 Neon today?
According to Google's blog, access is as follows:
- Gemini app: AI Pro and AI Ultra subscribers can pick Neon from the model dropdown.
- Gemini API: available in Google AI Studio, with Android Studio also mentioned for developers.
- Gemini for business: available through the API and the model dropdown.
- Google Antigravity: available to paid AI Pro and Ultra subscribers.
Android Authority notes that Carbon was limited to AI Ultra subscribers, while Neon is also open to AI Pro. Neither Google's post nor the coverage we read gives per-plan usage limits, so we cannot say how many Neon prompts a subscriber gets.
What is Gemini 4 Neon for cyber?
Google says a cybersecurity variant, Gemini 4 Neon for cyber, will roll out "in the coming days" to trusted defenders through its Fairwind Program, with requests routed through the Cloud account team. Google cites results on security benchmarks such as SEC-bench Pro (90.5%) and says it produces more secure code than Carbon. These are Google's claims and we have not seen independent testing. This is the stage of the rollout to watch, and it is separate from the general release.
Analysis: why the naming and the pricing matter
This section is our analysis, not reporting.
Neon is the fourth name in the Gemini 4 family, after Hydrogen, Argon and Carbon. A top comment under the 9to5Google story already complains that the old Flash and Pro labels were easier to follow. Google's answer is the plain-language model picker, which attaches a task to each name.
The more important change is access. A Carbon-adjacent model on AI Pro puts the strongest tier of Google's reasoning in front of far more paying users than Carbon itself reached. If the cost-per-task claim holds up in independent tests, developers have a reason to default to Neon and reserve Carbon for the hardest jobs. Google also released its Genie 4 world model to AI Ultra users on 18 November, so subscriber tiers are clearly a central part of its strategy this week.
Readers should keep one caution in mind. The benchmark table compares Google's own settings with rival models, and leaderboard positions change quickly. Wait for independent evaluations before treating the rankings as settled.
Sources
FAQ
Frequently asked questions
Google announced Gemini 4 Neon on 18 November 2026, and both its blog post and model card carry that date. Some outlets updated their coverage on 19 November, which is why you may see the later date.
Google lists $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.
On Google's own comparison table Neon leads on most benchmarks, but Claude Opus 5.5 is slightly ahead on two coding tests and GPT-6.2 Sol is ahead on one chart-reading test. These are Google's published results, not independent ones.
No. Google says AI Pro and AI Ultra subscribers can select Neon in the Gemini app. Developers can use it through the Gemini API in Google AI Studio.
Google positions Carbon for the hardest long-horizon projects and Neon as a balance of depth and efficiency. The model card says Neon is adapted from Carbon and Google claims it costs 32% less per task at the default setting.
Spotted an error? Tell us. Read how we report and correct stories in our editorial policy.
Written by
Mainstream Tech
Newsroom
News and analysis on technology, AI, business, politics and international affairs, from our newsroom in London.
More from Mainstream




