Quick answer
To try Gemini 4 Neon in Google AI Studio, sign in at aistudio.google.com, choose Gemini 4 Neon from the model list, and start on the default “high” thinking level. You need a Google AI Pro or Ultra plan, or a Gemini API key on the paid tier, because Neon is not on the free API tier. Drop to low or medium for quick, simple jobs, and save xhigh for hard problems where you can wait.
Key takeaways
- Neon is paid-only on the Gemini API: $3 per million input tokens and $15 per million output tokens.
- AI Pro and Ultra subscribers can try it in AI Studio, the Gemini app and Antigravity.
- Start at “high”, Google’s default, then move up or down based on results.
- xhigh can be slow: one 3D test in WorldofAI’s review took about 39 minutes.
📘 Main article: This guide follows on from our Gemini 4 Neon Review: What the Tests Show on Speed, Cost and Quality, which covers the benchmarks, pricing and hands-on test results.
Gemini 4 Neon launched on 18 November 2026, and early tests suggest it is close to the top models while being faster and cheaper per task. If you read our Gemini 4 Neon review and want to see for yourself, this guide shows the quickest way in and how to choose a thinking level without wasting money.
Where can you use Gemini 4 Neon?
You can use Neon in four places, depending on your plan. According to Google’s announcement:
| Where | Who can use it | Best for |
|---|---|---|
| Gemini app | Google AI Pro and Ultra subscribers | Quick chats, writing, simple builds |
| Google AI Studio | AI Pro and Ultra subscribers, or anyone with a paid Gemini API key | Testing prompts and comparing thinking levels |
| Google Antigravity | Paid AI Pro and Ultra subscribers | Multi-step coding and agent work |
| Gemini API | Paid tier only | Your own apps and automations |
⚠️ Warning: Neon is not available on the free Gemini API tier. If a site offers you “free Gemini 4 Neon access”, be wary, especially if it asks for your Google login or API key.
How do you try Gemini 4 Neon in Google AI Studio?
It takes about five minutes. Menu names in AI Studio change from time to time, so look for the closest match if a label differs.
- Sign in at aistudio.google.com with the Google account that holds your AI Pro or Ultra plan.
- Start a new chat or prompt.
- Open the model picker and choose Gemini 4 Neon (its API name is
gemini-4-neon). - Find the thinking level setting in the run settings panel. Leave it on high for your first try.
- Paste a real task you would normally give an AI tool, not a trick question.
- Note the time and token count shown for the response. You will use these to estimate costs.
- Run the same prompt at medium and compare. If the answer is just as good, you have found a cheaper setting for that job.
💡 Tip: If you plan to use the API, set a budget alert in your Google Cloud billing account before you start. It is the simplest way to avoid a surprise bill while you experiment.
Which thinking level should you choose?
Match the level to how hard the task is and how long you can wait. Higher levels think longer, and thinking tokens are billed as output, so they cost more.
| Level | Use it for | Watch out for |
|---|---|---|
| Low | Short rewrites, summaries, simple formatting, quick answers | Can miss details on multi-step tasks |
| Medium | Everyday writing, emails, simple code changes | Check logic on anything complex |
| High (default) | Most real work: reports, landing pages, planning, coding | A good balance; start here |
| xhigh | Hard engineering, analysis or science problems | Slow and costs the most; one demo took about 39 minutes |
For a deeper explanation of effort levels across different AI tools, see Low, High or Max: How to Choose an AI Reasoning Effort Level.
How do you estimate what a task will cost?
Multiply the tokens a task uses by Neon’s prices. AI Studio shows the token counts for each response.
| Example task | Input tokens | Output tokens | Approximate cost |
|---|---|---|---|
| Rewrite a 1,000-word article | 2,000 | 3,000 | About $0.05 |
| Summarise a 50-page report | 40,000 | 2,000 | About $0.15 |
| Build a landing page section (with thinking) | 5,000 | 20,000 | About $0.32 |
These are illustrative sums using Google’s published prices ($3 per million input, $15 per million output), not measured results. Your real numbers will depend on your prompt and the thinking level. Repeated input that hits the cache is billed at only $0.15 per million.
What should your first test be?
Pick one task you already do with AI and compare Neon with your current tool. A fair test looks like this:
- Use the exact same prompt in both tools.
- Score each result out of 5 for quality, and note how long it took.
- Check any facts, figures or links in the answer.
- Work out the rough cost of each using the method above.
- Decide: switch this task to Neon, keep your current model, or test again with a different thinking level.
Your next step: open AI Studio, choose Gemini 4 Neon on “high”, and run one task you did with AI this week. Then run it again on “medium” and see whether you can tell the difference.
If Neon gives you a strong first draft but not a finished result, our guide How to Split a Project Between a Fast, Cheap AI Model and a Premium One shows how to pair it with a premium model.
Frequently asked questions
Is Gemini 4 Neon free in AI Studio?
Not on the free API tier. Google AI Pro and Ultra subscribers can try it in AI Studio, and API users need the paid tier.
What is the default thinking level for Gemini 4 Neon?
High. The four levels are low, medium, high and xhigh.
What is the model ID for the API?
gemini-4-neon.
Does a higher thinking level always give better answers?
No. For simple tasks it often makes little difference, while costing more and taking longer. Test before you settle on a level.
Can I use Neon in the Gemini app instead?
Yes, if you have Google AI Pro or Ultra. Choose it from the model drop-down.
The bottom line
Trying Gemini 4 Neon is quick if you have the right plan. Start on high, test one real task, then try medium to see if you can save money. Keep xhigh for the rare jobs that truly need it, and keep a note of what works so you can choose the right setting next time.
Related posts
- Gemini 4 Neon Review: What the Tests Show on Speed, Cost and Quality
- How to Compare AI Models on Token Cost, Not Just Output Quality
- Gemini 4 Hydrogen: Google’s Fast New Flash Model, Anthropic’s “Slow Down” Pledge and This Week’s AI News
