# How to Use Gemini 4 Neon: Pricing, Limits and Availability

> Google released Gemini 4 Neon on November 18, 2026. See who gets it, the API price per million tokens, the model ID, context limits and how to call it.

Source: https://www.epifive.com/blog/gemini-4-neon-use-pricing-availability · Published 2026-11-19 · EpiFive

## Key takeaways

- Google announced Gemini 4 Neon on November 18, 2026, and the Gemini API changelog lists it as generally available.
- The API model ID is gemini-4-neon, with a 1,048,576-token context window and text, image, audio, video and PDF input.
- Standard paid-tier pricing is $3.00 per 1M input tokens and $15.00 per 1M output tokens, with no Free tier.
- In the Gemini app, Google says Pro and Ultra subscribers can pick Neon from the model dropdown.
- Thinking has four levels (low, medium, high, xhigh), set through thinking_level in generation_config.

Google announced Gemini 4 Neon on November 18, 2026, as a model for complex software engineering, knowledge work and science workflows. Pro and Ultra subscribers can select it from the model dropdown in the Gemini app, and developers can call it in the Gemini API on the paid tier with the model ID `gemini-4-neon`. The quickest way to start is to open Google AI Studio with the model preselected, or to pick Neon in the Gemini app's model dropdown.

## What Gemini 4 Neon is

Gemini 4 Neon is a model in the Gemini 4 family. Google's announcement describes it as aimed at "complex software engineering, knowledge work, and science workflows" with a focus on fast, end-to-end task completion. The Gemini API changelog entry dated November 18, 2026 describes it as generally available, with a unified 1 million-token context window and four thinking levels.

According to Google's models page, Neon takes text, image, video, audio and PDF input and returns text only. It does not generate images or audio and is not supported in the Live API. The supported features listed on that page are:

- Code execution and computer use
- Function calling and structured outputs
- Grounding with Google Search and with Google Maps
- URL context and file search
- Context caching (implicit only; explicit caching is not supported)
- Agentic video understanding (defaults to static)

Google's announcement also says Neon uses the Gemini 4 safety and security stack, with real-time safeguards against cyber and biosecurity misuse. This guide leaves out performance rankings; Google's announcement cites third-party indexes, and you should judge the model on your own tasks.

## Who it is for and availability

The table below combines the announcement with Google's models and pricing pages, as read on November 19, 2026.

| Product | Who gets it | Notes from Google |
|---|---|---|
| Gemini API | Paid tier | Not offered on the Free tier. Billed per token. |
| Google AI Studio | Paid API use; all Pro and Ultra subscribers can try it | The announcement links `model=gemini-4-neon`. |
| Gemini app | Pro and Ultra subscribers | Select it from the model dropdown. Gemini Tasks can also run on Neon. |
| Google Antigravity | Paid Google AI Pro and Ultra subscribers | Listed in the announcement. |
| Gemini for business | Via the API and the model dropdown | Listed in the announcement. |
| Android Studio | Listed as available | The announcement names Android Studio without further detail. |
| Gemini 4 Neon for cyber | Trusted defenders in the Fairwind Program | Rolling out over a few days; interested organizations are told to contact their Cloud account team. |

### API pricing

Prices are per 1 million tokens in USD on the paid tier, from Google's pricing page.

| Tier | Input | Output (including thinking) | Context cache |
|---|---|---|---|
| Standard | $3.00 | $15.00 | $0.15 |
| Batch | $1.50 | $7.50 | $0.075 |
| Flex | $1.50 | $7.50 | $0.075 |
| Priority | $5.40 | $27.00 | $0.27 |

Google lists no separate cache storage fee for this model. The announcement describes the $0.15 cached input price as a 95% discount. Google has not published an Indian rupee price for the API on the pages we opened, so convert from USD at your bank's rate and add any taxes that apply to you.

### Regions

The announcement and models page do not list country restrictions for the API or the Gemini app. They do not say whether every plan in every country sees the model on day one, so check your own account. If you do not see Neon yet, a staged rollout is one possible reason, but Google has not said that in the pages we opened.

## Requirements

- **For the Gemini app:** a Google AI Pro or Ultra subscription, according to the announcement. Google's help page says manual model selection requires signing in.
- **For the API:** a Gemini API key from Google AI Studio and a billing-enabled (paid) project, since the model is not on the Free tier. Google's quickstart says new users get a project and key automatically and can click **Create API key** to make another.
- **For the SDK:** the Python package from `pip install -U google-genai`, per the quickstart.
- **For Gemini for business:** the announcement says it is available there; admin controls for it were not on the pages we opened.

## How to use Gemini 4 Neon, step by step

### In the Gemini app

1. Sign in at gemini.google.com with an account that has a Pro or Ultra subscription.
2. Open the model dropdown and select Gemini 4 Neon.

As of November 19, 2026, Google had not published step-by-step instructions with exact menu labels for Neon. The announcement says only that Pro and Ultra subscribers "can select it from the model dropdown." Google's Gemini Apps help page says the app uses auto model selection by default and that "if you want to manually select a different available model, you can select it from the menu." Steps 1 and 2 follow those statements only; the exact label may differ.

### In Google AI Studio

1. Sign in to Google AI Studio.
2. Open the link from the announcement, `https://aistudio.google.com/prompts/new_chat?model=gemini-4-neon`, which starts a chat with the model preselected.
3. Write your prompt and run it.

### In the Gemini API

1. Create an API key in Google AI Studio and export it: `export GEMINI_API_KEY="YOUR_API_KEY"`.
2. Install the SDK with `pip install -U google-genai`.
3. Call the model. Google's thinking documentation shows the following request, which we changed only by swapping the model ID for `gemini-4-neon` (the official sample names `gemini-4-hydrogen`). It uses the Interactions API:

```python
from google import genai

client = genai.Client()

interaction = client.interactions.create(
    model="gemini-4-neon",
    input="Audit this payment retry queue for race conditions and rewrite the lock acquisition logic.",
    generation_config={
        "thinking_level": "xhigh"
    }
)

print(interaction.output_text)
```

The REST equivalent posts to `https://generativelanguage.googleapis.com/v1beta/interactions` with the header `x-goog-api-key: $GEMINI_API_KEY`. Google's quickstart REST example also sends an `Api-Revision: 2026-05-20` header; check the current docs for the value before you copy it.

### Choosing a thinking level

Google's thinking page says `thinking_level` goes inside `generation_config` and that Neon accepts `low`, `medium`, `high` and `xhigh`, with `high` as the default. The announcement calls xhigh the deepest reasoning setting for the hardest workloads. Output tokens, including thinking, are billed at the output rate, so higher levels can cost more per request; the pages we opened give no token counts per level.

### Lowering cost

Google's models page says Neon supports the Batch API, Flex inference and Priority inference, with the prices above. Implicit context caching is supported; explicit caching is not.

## Limitations

- **Output cap:** 65,536 tokens by default for synchronous requests. Google says its long decode feature can raise output to 1,048,576 tokens. We did not open the long-decode page.
- **Shared window:** the 1,048,576-token window is shared by input and output.
- **No Free tier** in the API.
- **Text output only:** no image or audio generation, and no Live API.
- **Rate limits:** neither the announcement nor the models page lists rate limits or usage quotas. Check your project's limits in AI Studio.
- **Knowledge cutoff:** not stated on the pages we opened.
- **Retirement of older models:** the pages we opened do not name a model that Neon replaces.
- **Gemini app limits:** the announcement gives no per-plan usage caps for Neon.

## Privacy and admin notes

The pages we opened for this guide do not state whether prompts to Neon are used for training. Review the Gemini API terms and the Gemini Apps privacy documentation for your account type before sending sensitive data. In the Gemini API, billing is a paid-tier setting, which is the tier Google's pricing page ties to this model.

For organizations, the announcement says Neon is available in Gemini for business through the API and the model dropdown, but it does not describe admin settings to allow or block it. The restricted cyber variant is limited to trusted defenders in the Fairwind Program; organizations are directed to their Cloud account team.

If your team also uses GitHub Copilot, note that model lists there change separately; see our guide to [GitHub Copilot model deprecations on December 14, 2026](https://www.epifive.com/blog/github-copilot-model-deprecation-december-2026).

## What we checked

- Opened Google's announcement, "Introducing Gemini 4 Neon," on 2026-11-19; it gives the November 18, 2026 date, the plan list, and the $3 / $0.15 / $15 prices.
- Opened the Gemini API models page on 2026-11-19; it confirms the model ID `gemini-4-neon`, the 1,048,576-token window, 65,536 default output, input and output modalities, and the four thinking levels.
- Cross-checked price: the announcement and the pricing page agree on $3.00 input, $15.00 output and $0.15 cached input (Standard). The Batch, Flex and Priority prices come only from the pricing page.
- Cross-checked the context window and thinking levels across the announcement, the models page and the changelog entry (all state 1M tokens and four levels).
- Opened the thinking documentation for the `thinking_level` parameter and sample request, and the quickstart for key setup and the SDK install command.
- Opened the Gemini Apps help page; it does not mention Neon or give steps for the model menu, so our app steps stay at the level of Google's announcement.
- Not tested: we did not call the API or use the Gemini app with a paid account, and we did not read the model card PDF in full (it did not load as readable text) or the long-decode page.

Last tested: 2026-11-19

## Sources

- [Introducing Gemini 4 Neon](https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-neon/)
- [Gemini API models](https://aistudio.google.com/docs/models) and [Gemini 4 Neon model page](https://aistudio.google.com/docs/models/gemini-4-neon)
- [Gemini API pricing](https://aistudio.google.com/docs/pricing)
- [Gemini API changelog](https://aistudio.google.com/docs/changelog)
- [Gemini API thinking guide](https://aistudio.google.com/docs/thinking)
- [Gemini API quickstart](https://aistudio.google.com/docs/quickstart)
- [Gemini Apps help: choose a different Gemini model](https://support.google.com/gemini/answer/14517446)

## Frequently asked questions

### How much does Gemini 4 Neon cost in the Gemini API?

On the Standard paid tier, Google's pricing page lists $3.00 per 1M input tokens, $15.00 per 1M output tokens (including thinking) and $0.15 per 1M cached input tokens. Batch and Flex are $1.50 input and $7.50 output; Priority is $5.40 input and $27.00 output.

### Is Gemini 4 Neon free?

Not in the API. Google's pricing page says the model is not offered on the Free tier. In the Gemini app, Google's announcement says Pro and Ultra subscribers can select it from the model dropdown.

### What is the Gemini 4 Neon model ID?

The model ID is gemini-4-neon. Google's models page lists no dated variants for it.

### How big is the context window?

The context window is 1,048,576 tokens, shared between input and output. Synchronous output is capped at 65,536 tokens by default; Google says long decode can raise output to 1,048,576 tokens.

### Can I turn off or lower the thinking?

Neon supports four thinking levels: low, medium, high and xhigh, with high as the default per Google's thinking documentation. Set thinking_level in generation_config to change it.

