Gemini 4 Neon: Google's new mid-tier model, explained
By the ChatGBot Team · Published November 20, 2026
Google has released Gemini 4 Neon, the third Gemini 4 model in less than a month. It arrived on November 18, 2026 and sits between the fast Gemini 4 Hydrogen and the most powerful Gemini 4 Carbon. Google says it brings frontier-level results at a lower cost, and on its own table it beats Claude Opus 5.5 and GPT-6.2 Sol on most tests. Here is what Neon is, how it scores, what it costs and who should use it.
Every top AI in one app
GPT, Claude, DeepSeek, Qwen and GLM. One subscription.
Try ChatGBot →Gemini 4 Neon at a glance
| Gemini 4 Neon | |
|---|---|
| Made by | |
| Released | November 18, 2026 |
| Model ID | gemini-4-neon |
| Built from | Gemini 4 Carbon (per the model card) |
| Context window | 1,048,576 tokens |
| Max output | 65,536 tokens per response, more with long decode |
| Input | Text, images, audio, video, PDF |
| Thinking levels | Low, medium, high (default), xhigh |
| API price | $3 input / $15 output per 1M tokens |
| In the Gemini app | "Deeper analysis", Google AI Pro and Ultra |
What is Gemini 4 Neon?
The Gemini 4 family now has three main models. Carbon is for the hardest, longest projects. Hydrogen is for fast, everyday work. Neon is the middle option: Google says it "balances depth and efficiency" and built it for complex coding, analysis and enterprise workflows. According to the model card, Neon was adapted from Carbon, so it shares Carbon's design but runs faster.
In the Gemini app, Neon now appears under "Deeper analysis", where Gemini 4 Argon used to be. Google's developer guide also lists Neon as the upgrade path from Argon, Gemini 3.1 Pro and Gemini 3.8 Flash.
What Neon is good at
- Coding. Google says Neon handles repo-level debugging and multi-file changes in fewer steps. It scores 65.4% on CursorBench 4.1 and 65.2% on Terminal-bench 4.1.
- Agent work. It scores 56.3% on AutomationBench and 75.8% on Agents' Last Exam v1.1, ahead of Opus 5.5 and GPT-6.2 Sol on both.
- Very long documents. On GraphWalks, a test across 256K to 1M tokens of context, it scores 95.2%, far ahead of Opus 5.5 (66.8%) and GPT-6.2 Sol (80.1%).
- Science and math. It scores 84% on RiemannBench and 71.3% on HLE-Diamond.
- Office files. On Google Cloud it reads Word, PowerPoint and Excel files directly.
Benchmarks
| Test | Gemini 4 Neon | Claude Opus 5.5 | GPT-6.2 Sol |
|---|---|---|---|
| Artificial Analysis Intelligence Index | 61 | 59 | 57 |
| Vals Index v2.2 | 70.9% | 70.1% | 66.1% |
| AutomationBench | 56.3% | 42.5% | 41.3% |
| Agents' Last Exam v1.1 | 75.8% | 71.6% | 73.0% |
| GraphWalks (256K to 1M) | 95.2% | 66.8% | 80.1% |
| Terminal-bench 4.1 | 65.2% | 65.8% | 61.5% |
| FrontierCode 2.0 | 46.2% | 47.3% | 43.8% |
| HLE-Diamond | 71.3% | 67.9% | 64.0% |
| Chartography | 73.5% | 66.3% | 75.1% |
These numbers come from Google. Neon trails Opus 5.5 on Terminal-bench 4.1 and FrontierCode 2.0, and trails GPT-6.2 Sol on Chartography. On the independent Artificial Analysis ranking, Neon places third, behind GPT-6.2 Astra and Claude Fable 5.2, which Google left out of its table. For more on OpenAI's side, see our guides to GPT-6.2 Sol and GPT-6.2 Astra.
Neon vs Hydrogen vs Carbon
| Hydrogen | Neon | Carbon | |
|---|---|---|---|
| Gemini app label | Everyday help | Deeper analysis | Most complex work |
| Best for | Fast answers and quick agent steps | Complex coding and knowledge work | The hardest long projects |
| API price per 1M tokens | $1.50 / $7.50 | $3 / $15 | $2 / $10 until Nov 30, then $4 / $20 |
| Artificial Analysis Index | 51 (high) | 61 (xhigh) | 60 |
Artificial Analysis found Neon about twice as fast as Carbon: around 110 output tokens per second versus 57, and about 21 minutes per index task versus 36. Its cost per task was about 17% lower than Carbon's. One caveat: Artificial Analysis also measured about 15% more output tokens per task than Carbon. Note that Carbon's intro price ends on November 30, so from December 1 Neon is the cheaper of the two per token.
Price and how to get access
- Gemini API and AI Studio: $3 per million input tokens and $15 per million output tokens, including thinking tokens, flat across the full context window. Cached input costs $0.15 per million. Batch and Flex modes cost half. Paid projects only.
- Gemini app: Google AI Pro and Ultra subscribers, under "Deeper analysis".
- Antigravity: Google AI Pro and Ultra subscribers. The Orchestrator mode can pair Neon for planning with Hydrogen for the edits.
- Businesses: through Gemini for business and Google Cloud.
Should you use it?
If you work on complex code, long documents or detailed analysis and want strong results without Carbon's wait, Neon is the Gemini 4 model to try. For quick everyday questions, Hydrogen is faster and cheaper. Results also vary by task, and Claude and GPT still lead on some tests, so it pays to compare models on your own work. For everything released this month, see our November 2026 AI model releases roundup.
Compare top models in one place
Ready to try it yourself?
Switch between top AI models in one chat.
Try ChatGBot →Frequently asked questions
What is Gemini 4 Neon?
Gemini 4 Neon is Google's mid-tier Gemini 4 model, released on November 18, 2026. It sits between the fast Gemini 4 Hydrogen and the most powerful Gemini 4 Carbon, and is built for complex coding, knowledge work and science tasks.
Does Gemini 4 Neon replace Gemini 4 Argon?
In the Gemini app, reporters saw Neon take Argon's place in the model picker under "Deeper analysis". Google's developer guide also names Neon as the upgrade path from Argon, Gemini 3.1 Pro and Gemini 3.8 Flash.
How much does Gemini 4 Neon cost?
In the Gemini API it costs $3 per million input tokens and $15 per million output tokens, flat across the full context window. Cached input costs $0.15 per million. There is no no-cost tier: Neon runs only on paid API projects.
Is Gemini 4 Neon better than Claude Opus 5.5 and GPT-6.2 Sol?
On Google's own benchmark table it leads both on most tests, including the Artificial Analysis Intelligence Index (61 vs 59 and 57). It trails Opus 5.5 on Terminal-bench 4.1 and FrontierCode 2.0, and GPT-6.2 Sol on Chartography.
What is the Gemini 4 Neon context window?
1,048,576 tokens. A single response is up to 65,536 tokens, and longer output is possible with long decode or background execution. It reads text, images, audio, video and PDFs.
Is Gemini 4 Neon available in ChatGBot?
No. ChatGBot offers GPT, Claude, DeepSeek, Qwen and GLM models in one app, so you can compare top models side by side.