Google Releases Gemini 4 Neon, 32% Cheaper Than Carbon
Neon sits between Carbon and Hydrogen in the Gemini 4 family. A million input tokens cost $3, and a million output tokens cost $15.
Google has released Gemini 4 Neon, a new artificial intelligence model. It sits in the middle of the Gemini 4 family. Above it is Carbon, which Google pitches for the longest projects. Below it is Hydrogen, meant for quick everyday tasks. Neon is available to developers through the Gemini API, the service programs use to call Google's models. People who pay for the Pro and Ultra plans can also use it in the Gemini app.
The main selling point is price. Models like this charge by the token, a small piece of text. Neon costs $3 per million tokens sent in and $15 per million tokens it writes back. Google says that at the standard reasoning setting, a typical task costs 32% less than on Carbon. For a company that runs thousands of tasks a day, that difference adds up.
Nearly every result in the announcement comes from Google itself. The company says Neon ranks third, with 61 points, on the Artificial Analysis index, a widely watched ranking of AI models. That score was reached at Neon's highest reasoning setting, and Neon still sits behind larger, slower models. So this is another attempt by Google to catch up with the leaders, this time by competing on price. Independent tests on real work will show whether the lower cost holds up.
Source: deepmind.google
Free course
Stop reading about AI — start building with it
The free Claude Code course: your first site, tool or game — no coding. No upsells, no cross-sells — nothing to buy here.
Start free →
Author
Evgenii Arsentev
PhD · Founder & CEO, Dad4Kids
Articles · Latest articles