On this page · 6 sections ▾
Google launched Gemini 4 Neon on November 18, 2026, a mid-tier model in the Gemini 4 family that sits between Carbon and Hydrogen. It is available to Google AI Pro and Ultra subscribers in the Gemini app, and to developers through the paid Gemini API. Neon replaces Gemini 4 Argon in the app's model picker, according to 9to5Google.
Key facts
- Positioning: Neon targets complex coding, analytical reasoning and production-grade enterprise work. Google says it matches or approaches Carbon at lower cost and higher speed.
- Pricing: $3 per million input tokens and $15 per million output tokens, with cached input at $0.15, per Digital Trends.
- Specs: a 1 million-token context window and four thinking levels (Low, Medium, High, Extra High), with High as the default.
- Intelligence Index: a score of 61 at Extra High, third overall on the Artificial Analysis Intelligence Index.
- Availability: the Gemini app, Google AI Studio, Android Studio, Gemini for business and Google Antigravity.
What changes in the Gemini app
9to5Google reports that the model picker now offers Auto, Everyday help (4 Hydrogen), Deeper analysis (4 Neon) and Code & complex work (4 Carbon). Gemini 3.1 Pro is being retired, and 3.5 Flash-Lite now powers "Auto" for paid users.
Google says Neon can run Tasks in Gemini, and that it reads spreadsheets, PDFs and photos and pulls details from Gmail and Docs. Examples include comparing vendor pricing and checking a utility bill for unusual charges.
Speed, cost and benchmarks
Google's claims differ somewhat by outlet. FoneArena reports that at the default High setting Neon costs 32% less per task than Carbon while keeping a high Intelligence Index score. India Today cites Google as saying Neon runs roughly three times faster than Carbon and cuts overall costs by about 60 percent. The 32% figure is per task, so the two numbers are not directly comparable.
Other reported results include 70.9% on the Vals Index, first place on the Legal Research Bench, Tax Agent Bench and Zapier's AutomationBench (56.3%), and 65.2% on Terminal-Bench 4.1. Neon also scored 46.2% on FrontierCode 2.0 at the High setting.
The competitive picture depends on the comparison. Digital Trends says Google's table puts Neon at 61 on the Intelligence Index, ahead of Claude Opus 5.5 (59) and GPT-6.2 Sol (57), but behind Opus 5.5 on Terminal-Bench 4.1 and FrontierCode 2.0. India Today, citing Artificial Analysis, places it behind Claude Fable 5.2 (65) and GPT-6.2 Astra (62), and ahead of Carbon (60). Most of these figures are supplied by Google, and Digital Trends notes that independent tests may differ.
Pricing catch
Both 9to5Google and India Today report that Argon and Carbon are priced at $2 input and $10 output until November 30. Carbon then moves to $4 and $20, which would make Neon 25% cheaper than Carbon from December 1. Until then, Neon costs more per token than the promotional Carbon rate, though Google's per-task figures point the other way.
India Today also places Neon above GPT-6.2 Sol on input price ($2) and below Claude Opus 5.5 ($4).




