Google launches Gemini 4 Neon, a mid-priced AI model aimed at coding and enterprise work
Google's fourth Gemini 4 model in under two months lands between Hydrogen and Carbon at $3 and $15 per million tokens, pressing rivals on coding performance and price.
By Hushread Stories, from 42 outlets · First published 20 Nov 2026 · Chapter 2 of 2
Earlier: Google launches Gemini 4 Neon, a cheaper mid-tier model aimed at coding and agents
In brief
- Google released Gemini 4 Neon on November 18, the fourth Gemini 4 model in under two months.
- Neon costs $3 per million input tokens and $15 per million output tokens, with a 1 million-token context window.
- It sits between the fast Hydrogen model and the more powerful Carbon, and targets coding, agents and enterprise workloads.
- Google AI Pro and Ultra subscribers get Neon in the Gemini app, replacing Argon in the model picker.
- Outlets and reviewers are already comparing it with Claude Opus 5.5 and GPT-6.2 Sol, with some noting confusing pricing.
Timeline · 9 moments
Gemini for Mac gains computer use
TechCrunch ↗Genie 4 opens to developers via Gemini API
TestingCatalog ↗Reviewer tests Neon against Opus 5.5 and Carbon
Flavio Copes ↗Neon reaches Pro and Ultra users, replacing Argon
Digital Trends ↗Genie 4 opens to AI Ultra subscribers
The Next Web ↗Gemini Tasks becomes free in six countries
Digital Trends ↗Neon ranks second on Agent Arena, third on AA Index
TestingCatalog ↗Google announces Gemini 4 Neon with $3/$15 pricing
BigGo Finance — Real-time Stock Quotes, Financial & Business News ↗Google launches Gemini 4 Hydrogen
Tom's Guide ↗How it started
Google has been rolling out its Gemini 4 family at a fast pace. 9to5Mac noted that Neon arrived after the release of Carbon and Hydrogen the previous week, and Tom's Guide said Hydrogen launched on November 12.
The result was a lineup with a gap in the middle. As Habr put it, users had to choose between the fast Hydrogen model for simple scripts and routine work and the powerful but expensive Carbon.
How it unfolded
On November 18, Google announced Gemini 4 Neon, with a post on its own blog and a Hacker News thread that gathered more than 500 points. VentureBeat reported that it launched on the paid Gemini API in Google AI Studio and Android Studio, as well as in the Gemini app and Google's AI-assisted development platform.
BigGo Finance reported the pricing of $3 per million input tokens and $15 per million output tokens, plus a 1 million-token context window, and said Google ranks the model third overall on the Artificial Analysis index. TestingCatalog added that Google places it second on Agent Arena, and that it is available in Gemini, AI Studio and Antigravity.
The headline claim came from VentureBeat, which described Opus 5.5-level coding performance at 25% lower API token prices. Technology reported that Google says Neon is three times faster than Carbon and can beat it in some benchmarks, though it called the pricing situation a little unusual. InfoWorld framed the release as part of Google's push to expand the use of AI coding agents, with the company saying Neon can finish complex software engineering work more economically than larger models.
On November 19, Digital Trends reported that Neon is live for Google AI Pro and Ultra subscribers and replaces Argon in the model picker. The same day, early testing appeared. Flavio Copes ran Neon against Opus 5.5 and Gemini 4 Carbon on coding, writing and search tasks, tracking real costs.
Neon was one of several Google announcements that week. The Next Web reported Genie 4 reaching AI Ultra subscribers, and TestingCatalog said it opened to developers in the Gemini API. Digital Trends covered the Gemini Tasks feature going free in six countries, and TechCrunch reported Gemini gaining computer use on the Mac.
Where it stands
Neon is live across the Gemini app, the paid API and Google's developer tools. Coverage broadly agrees on its pitch: frontier-level reasoning at a lower cost and higher speed than Carbon.
Not everyone is celebrating the packaging. Simon Willison wrote that the Gemini 4 pricing was already confusing and that Neon made it worse. A French outlet, tech, said the rapid run of models comes from Google's wish to segment uses and cut costs, at the risk of confusing users. Most performance claims so far come from Google or from early hands-on tests.
What to watch
The main open question is how Neon holds up in independent testing against Opus 5.5 and GPT-6.2 Sol, since the strongest numbers currently come from Google's own positioning. Developers will also be weighing the real cost of long agentic runs against the headline token prices.
Another is whether Google keeps adding models to the Gemini 4 family at this pace, and whether it can make the tiers and prices easier to understand.


