What Is Gemini 4 Neon? Price, Benchmarks and Access
The short answer
Gemini 4 Neon is Google’s mid-tier Gemini 4 model, released November 18, 2026: near-Carbon quality for coding and analysis at $3/$15 per million tokens, with Hydrogen-like speed. Google positions it between Hydrogen (fast everyday execution) and Carbon (the hardest projects).
| Gemini 4 Hydrogen | Gemini 4 Neon | Gemini 4 Carbon | |
|---|---|---|---|
| Model ID | gemini-4-hydrogen | gemini-4-neon | gemini-4-carbon |
| Released | Nov 12, 2026 | Nov 18, 2026 | Nov 10, 2026 (to developers) |
| Input / cached / output (per MTok) | $1.50 / $0.075 / $7.50 | $3 / $0.15 / $15 | $2 / $0.10 / $10 through Nov 30 → $4 / $0.20 / $20 from Dec 1, 2026 |
| Free API tier | ✅ | ❌ | ❌ |
| Context | 1M | 1M, long decode up to 1M | 1M |
| Role | Fast everyday work | Complex coding and enterprise workflows | Hardest long-horizon projects |
Prices from Google’s Gemini API pricing page, November 19, 2026. Google lists no introductory period for Neon. Note the timing: until November 30, Carbon’s introductory price is below Neon’s; from December 1, Neon becomes the cheaper of the two. For all models, see our current API prices.
Pricing detail
- Standard: $3 input, $0.15 cached input (a 95% discount; implicit caching only, no explicit caches), $15 output including thinking tokens.
- Batch and Flex: $1.50 / $0.075 / $7.50.
- Priority: $5.40 / $0.27 / $27.
- Grounding with Google Search or Maps: 5,000 free requests a month shared across Gemini 3 and newer models, then $14 per 1,000 (Search capped at 1,500 a day).
- Data: paid-tier prompts are not used to improve Google’s products.
Benchmarks (Google’s announcement)
- Artificial Analysis Intelligence Index: 61 at xhigh, which Google says ranks it #3 overall behind larger, slower models.
- Artificial Analysis Coding Agent Index: 58.6, #1 per Google.
- Terminal-Bench 4.1: 65.2%. FrontierCode 2.0: 46.2% at the default high setting.
- Vals Index: #2 at 70.9%. Zapier AutomationBench: #1 at 56.3%. Agents’ Last Exam v1.1: 75.8%.
- Terminal-Bench Science: 70.5%.
- Cost: 32% lower cost per task than Carbon at high; Google claims 50–80% lower cost per task than top competing frontier models on Vals and Agent Arena.
These are vendor-reported. For coding, compare on your own tasks against GPT-6.2 Sol ($2/$10) and Claude Opus 5.5 ($4/$20).
Where you can use it
- Developers: Gemini API paid tier in Google AI Studio and Android Studio. Not on the free tier.
- Businesses: Gemini for business via API and the model drop-down, and Google Antigravity, where Neon can orchestrate parallel Hydrogen subagents.
- Consumers: Google AI Pro and Ultra subscribers can pick Neon in the Gemini app; paid Pro and Ultra subscribers also get it in Antigravity. See Gemini Free vs AI Plus vs AI Pro vs AI Ultra.
- Security teams: Gemini 4 Neon for cyber rolls out to Fairwind Program members.
Which Gemini 4 model to use
- Hydrogen for high-volume, latency-sensitive calls and anything that must run on the free tier.
- Neon as the default for coding agents, analysis and enterprise workflows from December 1, when Carbon’s price doubles.
- Carbon when a task fails on Neon and the extra cost is worth it, or through November 30 while its introductory price is lower.
Last verified: November 19, 2026.