AI and machine learning

Google releases Gemini 4 Neon, its 4th model in 7 weeks

Google DeepMind’s Gemini 4 Neon scores 61 on the Artificial Analysis Intelligence Index, third overall, at $3 per million input tokens and $15 per million output tokens.

Filed under Large language models, AI industry

Google released Gemini 4 Neon in its API, the Antigravity coding environment and the Gemini app on 18 November, aimed at software engineering, knowledge work and science. It has a context window of 1m tokens and costs $3 per million input tokens and $15 per million output tokens, 25% less than Gemini 4 Carbon.

Artificial Analysis, an independent benchmarking firm, scored Neon at 61 on its Intelligence Index at the highest thinking setting, one point ahead of Carbon and third overall behind GPT-6.2 Astra and Claude Fable 5.2. Its cost per task was about $6.44, roughly 17% below Carbon, and it finished tasks in about 21 minutes instead of 36. The firm called it Google’s fourth model in seven weeks.

Google said Neon ranks second on the Vals Index and second for steerability on the new Agent Arena Index, and recorded its lowest fabrication rate yet in internal testing.