Google DeepMind has introduced Gemini 4 Neon, the newest member of the Gemini 4 family, built to pair frontier-level intelligence with fast, end-to-end task completion across complex software engineering, knowledge work and scientific workflows.

On independent benchmarks, Neon sits just behind larger, slower models: it ranks third overall with a score of 61 on the Artificial Analysis Intelligence Index at its highest thinking level, second on the Vals Index, and second for steerability on the Agent Arena Index. For coding work it leads the Artificial Analysis Coding Agent Index, and it tops several specialist benchmarks covering legal research, tax work, finance and spreadsheet modelling.

Google is pitching Neon as the middle of the Gemini 4 range. Carbon remains the pick for the hardest, long-horizon projects and Hydrogen handles fast everyday execution, while Neon is meant for complex, high-frequency work where depth matters but cost and speed do too. At its default high thinking level, Google says Neon costs 32 percent less per task than Carbon, with four thinking levels that let teams trade intelligence against token spend.

Pricing is set at 3 dollars per million input tokens, with cached input at 15 cents per million, and 15 dollars per million output tokens, with a 1 million-token context window. A dedicated cyber variant, Gemini 4 Neon for cyber, is aimed at defenders: Google says it writes more secure code than Carbon and leads on vulnerability finding and patching benchmarks, and it is rolling out to trusted defenders through the Fairwind Program over the coming days.

Neon begins rolling out today in the Gemini API and Google AI Studio, in Gemini for business, and for paid Google AI Pro and Ultra subscribers in Google Antigravity and the Gemini app.