Google Launches Gemini 4 Neon: Faster Frontier AI for Coding, Knowledge Work and Science

Google has launched Gemini 4 Neon, a new artificial intelligence model that pairs frontier-level reasoning with faster task completion for software engineering, knowledge work and science. The announcement came on Wednesday 18 November 2026 from Google DeepMind, and the model is rolling out the same day across the Gemini API, Google AI Studio and the Gemini app for paying subscribers.
Neon is the fourth model in Google’s Gemini 4 family, arriving six days after the fast, low-cost Hydrogen model and a little over a month after the flagship Argon model. Google positions Neon in the middle of the range: deeper than Hydrogen, cheaper and faster than the heavyweight Carbon model, and built for the complex, high-frequency work that developers and businesses run every day.
Gemini 4 Neon at a Glance
| Detail | Gemini 4 Neon |
|---|---|
| Announced | Wednesday 18 November 2026 |
| Announced by | Koray Kavukcuoglu, SVP Google DeepMind and Chief AI Architect |
| Built for | Software engineering, knowledge work and science workflows |
| Intelligence Index score | 61 at xhigh thinking level, third overall (Artificial Analysis) |
| Coding Agent Index | First overall at 58.6 (Artificial Analysis) |
| API pricing | $3 per million input tokens, $15 per million output tokens |
| Cached input price | $0.15 per million tokens, a 95% discount |
| Context window | 1 million tokens |
| Thinking levels | Low, medium, high (default) and xhigh |
| Available in | Gemini API, Google AI Studio, Android Studio, Gemini app, Gemini for business, Google Antigravity |
How Neon Fits Between Hydrogen and Carbon
The Gemini 4 family now has four members, each with a different job. Argon, generally available since 13 October 2026, is the flagship reasoning model for the hardest, long-horizon problems. Carbon, released on 10 November, supplies deeper reasoning inside Google Antigravity, the company’s agentic coding environment. Hydrogen, launched on 12 November, handles fast everyday execution, and our earlier report on Gemini 4 Hydrogen set out its speed-first pitch.
Neon sits between Hydrogen and Carbon. Google says it delivers frontier-level reasoning for complex, high-frequency workflows, faster and more cost efficiently than the larger models. In Antigravity it can also act as an Orchestrator, directing parallel Hydrogen subagents while keeping a task moving end to end.
Benchmarks Google Is Leading With
Google’s headline number is a score of 61 on the Artificial Analysis Intelligence Index at Neon’s highest, xhigh, thinking level, putting it third overall behind only larger, slower models. On the same provider’s Coding Agent Index, Neon ranks first overall at 58.6. It also ranks second on the Vals Index at 70.9%, and first on both the Legal Research Bench and the Tax Agent Bench.
The coding figures continue that pattern. Neon scores 65.2% on Terminal-Bench 4.1 and 46.2% on FrontierCode 2.0 at its default high setting. On scientific and mathematical reasoning, Google says Neon is best in class on Terminal-Bench Science at 70.5%, RiemannBench at 84.0% and HLE-Diamond at 71.3%. All of these figures are Google’s own claims from its announcement, and independent replication has not yet been published.
What Neon Costs, and the Carbon Price Twist
Neon is priced at $3 per million input tokens and $15 per million output tokens, with cached input at $0.15 per million. At its default high thinking level, Google says it costs 32% less per task than Carbon, and roughly 50% to 80% less per task than top competing frontier models.
There is a quirk in the pricing that buyers should know about. Carbon and Argon are on promotional pricing of $2 per million input tokens and $10 per million output tokens until 30 November 2026, according to 9to5Google’s coverage, after which Carbon rises to $4 and $20 on 1 December. Until that date, the flagship model is actually cheaper per token than Neon; after it, Neon becomes the lower-cost option by 25%.
Where Developers and Subscribers Can Use It
Neon began rolling out on 18 November 2026 in the Gemini API paid tier in Google AI Studio and Android Studio, in Gemini for business through the API and the model drop-down, in Google Antigravity for paid Google AI Pro and Ultra subscribers, and in the Gemini app for Pro and Ultra subscribers. Tasks in the Gemini app can now run on Neon, which Google says works through multi-step jobs more quickly than larger models.
A separate cyber variant, Gemini 4 Neon for cyber, is rolling out to trusted defenders in the coming days as part of Google’s Fairwind Program. Google says the cyber variant produces more secure code than Carbon across easy, medium and hard challenges on AutoBaxBench, reaching state of the art on the hard tier at 76%.
The Safety Picture Google Has Shared
Neon carries the same level of protection against severe misuse and prompt injection as Carbon, according to Google. The company says Neon recorded its lowest fabrication rate yet in internal Antigravity testing, and that at its highest thinking level it resisted every attack in Gray Swan’s indirect prompt injection benchmark, matching a 0.0% attack success rate.
Google also reports that Neon took shortcuts on just 2.45% of tasks in its coding cheating evaluation, the lowest rate of any Gemini 4 model, and matched Carbon with 0.0% cheating on Terminal-Bench 4.0. As with the benchmarks, these are Google’s internal figures, and the full model card is the place to check the detail.
What the Launch Means for the AI Race
Neon’s arrival tightens the contest for the everyday developer workload, and it underlines how quickly Google is shipping the Gemini 4 family: four models have reached general availability in barely five weeks. The pricing twist around Carbon’s expiring promotion also sets a clear date — 1 December — after which Neon becomes the rational default for teams that were running on the flagship.
Whether Neon wins developers over will depend on real-world reliability rather than launch-day benchmarks. For now, the most concrete change is the simplest one. From this week, a fourth Gemini 4 model is available in the tools developers already use, aimed squarely at the long, tool-heavy jobs that used to need the biggest model in the room. Google’s own announcement is the primary source for the figures in this article, and the full Gemini 4 Neon announcement is published on the Google blog.
Frequently Asked Questions
What is Gemini 4 Neon?
Gemini 4 Neon is Google’s fourth Gemini 4 AI model, announced on 18 November 2026. It is built for complex software engineering, knowledge work and science workflows, balancing frontier-level reasoning with fast, lower-cost task completion.
How much does Gemini 4 Neon cost?
Neon costs $3 per million input tokens and $15 per million output tokens, with cached input at $0.15 per million tokens. At its default high thinking level, Google says it costs 32% less per task than Carbon.
How does Gemini 4 Neon rank on benchmarks?
Google says Neon scores 61 on the Artificial Analysis Intelligence Index at its xhigh thinking level, third overall, and ranks first on the Artificial Analysis Coding Agent Index at 58.6. These are Google’s claims from its announcement.
Where can I use Gemini 4 Neon?
Neon is rolling out in the Gemini API in Google AI Studio and Android Studio, in Gemini for business, in Google Antigravity for paid Google AI Pro and Ultra subscribers, and in the Gemini app for Pro and Ultra subscribers.
How is Gemini 4 Neon different from Hydrogen and Carbon?
Hydrogen is the fast, low-cost execution model, Carbon is the heavyweight for deep, long-horizon reasoning, and Neon sits between them, built for complex, high-frequency work at a lower cost per task than Carbon.




2 Comments