Google DeepMind released Gemini 4 Neon on Nov 18, its fourth model in seven weeks. The new model sits between the Carbon and Hydrogen tiers in the Gemini 4 family and targets complex software engineering, knowledge work, and science workflows.
Google says Neon ranks third overall on the Artificial Analysis Intelligence Index with a score of 61 at its highest thinking level, ahead of Claude Opus 5.5 at 59 and GPT-6.2 Sol at 57, though it lands behind the largest and slowest frontier models. It places second on the Vals Index at 70.9 percent and takes the top spot on both the Legal Research Bench and the Tax Agent Bench. On Zapier's AutomationBench it leads at 56.3 percent, and it debuts second on Arena's new Agent Arena Index. That ranking remains a Google claim, since Neon was not yet listed on the public Artificial Analysis leaderboard when the launch was announced.
Efficiency is the pitch. DeepMind says Neon costs about 6.44 USD per task, roughly 17 percent less than Carbon and 50 to 80 percent below the top competing frontier models. It ships with a 1M-token context window and four adjustable thinking levels, and Google reports its lowest fabrication rate yet in internal Antigravity testing. Neon is rolling out to Google AI Pro and Ultra subscribers in the Gemini app, Google AI Studio, and Antigravity, with paid API access for developers and enterprises. For founders building on frontier models, a cheaper high-capability tier that undercuts rivals on cost per task is the headline worth watching.
