Google released Gemini 4 Neon on Wednesday, November 18, a middle tier between its fast Hydrogen model and the flagship Carbon. In the Gemini API it costs $3 per million input tokens and $15 per million output tokens, with a 1 million-token context window. Google says Neon scores 61 on the Artificial Analysis Intelligence Index at its highest thinking level, third overall; all benchmark figures here are Google’s own claims.
What Google released
Google pitches Neon for complex coding, analysis and science tasks that run often enough for cost and speed to matter. Its model card says Neon is adapted from Gemini 4 Carbon and has a knowledge cutoff of March 2026. The API model ID is gemini-4-neon.
Per Google’s documentation, Neon reads text, images, video, audio and PDFs and writes text. Input, thinking and output share one 1,048,576-token window; synchronous replies stop at 65,536 tokens, while background “long decode” requests can use the full window. There are four thinking levels, low, medium, high and xhigh, with high as the default. Function calling, code execution and computer use are supported.
Pricing and availability
The rates are flat across the context window, and output prices include thinking tokens. Implicitly cached input costs $0.15 per million tokens. Batch and Flex cost half, Priority $5.40 and $27, and there is no free API tier.
Neon is rolling out in the paid Gemini API, Android Studio and Gemini for business. Google AI Pro and Ultra subscribers get it in the Gemini app and Google Antigravity. A cyber-defense version reaches vetted members of Google’s Fairwind Program over the coming days.
Benchmarks and speed claims
Google gives no tokens-per-second figure and frames speed as cost per task: at the default high level, it says Neon costs 32 percent less per task than Carbon and roughly 50 to 80 percent less than top rival frontier models on the Vals Index and Agent Arena.
Against Claude Opus 5.5 and GPT-6.2 Sol, the model card has Neon leading on the Artificial Analysis index (61 against 59 and 57) and the Vals Index (70.9 percent against 70.1 and 66.1). Coding is closer: Opus 5.5 edges Neon on Terminal-Bench 4.1 (65.8 against 65.2 percent) and FrontierCode 2.0 (47.3 against 46.2 percent).
How Neon compares with Hydrogen, Argon, Carbon and GPT-6.2 Sol
The lineup shifts on December 1, when Carbon and Argon double to $4 and $20 per million tokens.
| Model | Input / output per 1M | Cached input | Default thinking |
|---|---|---|---|
| Gemini 4 Hydrogen | $1.50 / $7.50 | $0.075 | high |
| Gemini 4 Neon | $3 / $15 | $0.15 | high |
| Argon and Carbon, to Nov 30 | $2 / $10 | $0.10 | medium |
| Argon and Carbon, from Dec 1 | $4 / $20 | $0.20 | medium |
| GPT-6.2 Sol | $2 / $10 | $0.10 | n/a |
Take the example from our Gemini 4 Hydrogen story: an agent that reads 20 million tokens and writes 2 million a day. It costs $90 on Neon, $45 on Hydrogen, and $60 on Argon or Carbon today, rising to $120 in December. GPT-6.2 Sol, which OpenAI kept at $2 and $10 in its GPT-6.2 launch, also costs $60.
For two weeks Neon is the pricier option per token; from December it undercuts Carbon and Argon by 25 percent. These sums ignore how many tokens each model spends per task, which is where Google says Neon saves money.
What to watch next
- Independent results that confirm or adjust Google’s scores.
- The Carbon and Argon price rise on December 1.
- The Gemini 4 Neon Safety and Security Report, which the model card lists as coming soon.
Key takeaways
- Gemini 4 Neon, adapted from Carbon, launched on November 18.
- It costs $3/$15 per million input and output tokens, with $0.15 cached input and no free tier.
- It has a unified 1,048,576-token context window and four thinking levels.
- Google claims 61 on the Artificial Analysis index and 32 percent lower cost per task than Carbon.
- It is in the Gemini API, Gemini for business, Antigravity and the Gemini app for Pro and Ultra users.
Frequently asked questions
How much does Gemini 4 Neon cost?
The standard API price is $3 per million input tokens and $15 per million output tokens, including thinking tokens. Cached input costs $0.15 per million tokens, and Batch and Flex requests cost half.
Is Gemini 4 Neon free to use?
No. Neon is not on the free tier of the Gemini API. In the Gemini app it is available to Google AI Pro and Ultra subscribers through the model menu.
What is the difference between Gemini 4 Neon and Gemini 4 Carbon?
Carbon is Google’s most capable model for the hardest long-horizon problems. Neon, adapted from Carbon, targets frequent complex tasks at about 32 percent lower cost per task, Google says.




