AI

Google releases Gemini 4 Neon for coding and business work at $3 per million input tokens

Google released Gemini 4 Neon on Wednesday, November 18, and it is rolling out to developers, businesses and paying Google AI Pro and Ultra subscribers. Google pitches the model as a middle option in its Gemini 4 lineup: close to its biggest models on hard coding and analysis work, but faster and cheaper per task.

By Techieslite ·

An orange neon sign spelling Gemini 4 Neon next to a four-pointed star glows on a dark wall, the header image from Google's announcement

Where Neon fits in the Gemini 4 family

Neon joins Argon, Carbon and Hydrogen in the Gemini 4 family. Google says Carbon is meant for the hardest long-running projects and Hydrogen for fast everyday tasks, while Neon balances depth and efficiency for complex coding, analysis and business workflows.

Like Hydrogen, Neon has 4 thinking levels, so developers can trade reasoning depth for speed and token spend. Google says that at the default "high" level, Neon costs 32 percent less per task than Carbon.

What Google says it can do

Google's benchmark claims are all self-reported or drawn from third-party leaderboards it selected. The company says Neon scores 61 on the Artificial Analysis Intelligence Index at its top "xhigh" setting, which puts it third overall. It also reports:

Google says Neon can work across spreadsheets, PDFs, charts, audio and video in a single 1 million-token context window. It also says Neon had the lowest fabrication rate of any Google model so far in internal testing in Google Antigravity.

On safety, Google says Neon keeps Carbon-level protections against severe misuse and prompt injection. It also says Neon turns down little legitimate work in cyber and biosecurity, citing a refusal rate of 1.98 percent on Terminal-Bench 4.0. A version called Gemini 4 Neon for cyber is rolling out to eligible members of Google's Fairwind Program for security defenders over the next few days.

Price and availability

Neon costs $3 per million input tokens and $15 per million output tokens. Cached input costs $0.15 per million tokens, a 95 percent discount. For comparison, Hydrogen costs $1.50 per million input tokens and $7.50 per million output tokens.

Starting November 18, Neon is rolling out in these places:

Google also says tasks in Gemini, its feature for handing off multi-step jobs, can now run on Neon.

What it means for you

If you use the free Gemini app, nothing changes today. In the Gemini app, Neon is limited to Google AI Pro and Ultra subscribers, who can pick it from the model menu. Google suggests handing it tasks like analyzing billing statements or cross-checking receipts, and says running them on Neon stretches usage limits further than larger models do.

Developers paying for the Gemini API get a model that costs twice as much per token as Hydrogen but, by Google's figures, 32 percent less per task than Carbon at the default setting. If Hydrogen falls short on a coding or analysis task, Neon is the next step up to test. Treat the benchmark rankings as Google's claims until independent testers have spent more time with the model.

Sources