← AI Switchboard
MODELS · November 20, 2026
Nov 18

Google releases Gemini 4 Neon at $3 per million input tokens, between Carbon and Hydrogen, and claims a 32% lower cost per task than Carbon

By the AI Switchboard · researched and written with AI from the sources listed on this page, each point checked against its source before publishing · how we work · report a mistake

Neon is the third Gemini 4 model in a week. Its model card says biological and cyber critical capability levels cannot be ruled out.

Google released Gemini 4 Neon on 18 November, a model for complex coding, analysis and business work that sits between Carbon, its strongest model, and Hydrogen, its fast one. Neon costs $3 per million input tokens and $15 per million output tokens, with a 1 million-token context window, Google's announcement says. That is twice the $1.50 input price of Gemini 4 Hydrogen, which Google released on 12 November (Switchboard, 12 November).

Google gives no Carbon price in the post. Instead it claims Neon has “a 32% lower cost per task than Carbon” at its default high thinking level, and that at its highest thinking level it ranks #3 overall with a score of 61 on the Artificial Analysis Intelligence Index. It also reports 65.2% on Terminal-Bench 4.1, and says Neon can act as an orchestrator directing parallel Hydrogen subagents in its Antigravity coding tool.

The model card, published 18 November, gives a knowledge cutoff of March 2026. Its safety results are carried forward from Gemini 4 Argon, and for biological and cyber risks they say the critical capability level “cannot be ruled out”.

The Gemini 4 rollout so far, from the Switchboard's stories: on 29 October Carbon topped the Artificial Analysis index at 59 and went first to cyber defenders; on 11 November Google opened Carbon to paying developers; on 12 November came Hydrogen, and on 18 November Neon. By Google's own figures Neon's 61 on the index is above the 59 Carbon scored when the Switchboard reported it on 29 October.

  • Confirmed Gemini 4 Neon costs $3 per million input tokens and $15 per million output tokens, with a 1 million-token context window. Google
  • Claimed Google says Neon has a 32% lower cost per task than Carbon at its default high thinking level and scores 61 on the Artificial Analysis Intelligence Index, #3 overall. Google
  • Confirmed The model card, published 18 November, gives a March 2026 knowledge cutoff and says biological and cyber critical capability levels cannot be ruled out (carried forward from Argon). Google DeepMind model card

Models & releases

Catching up in the November 20, 2026 edition · front page