What happened
Google DeepMind on Wednesday introduced Gemini 4 Neon, its fourth Gemini 4 release in roughly seven weeks, Trending Topics reported. Neon scored 61 points on Artificial Analysis's Intelligence Index when run at its top xhigh reasoning setting. That put it in third place, with Anthropic's Claude Fable 5.2 and OpenAI's GPT-6.2 Astra ahead, each tested at maximum effort. Google is positioning the model for demanding, repeated coding, reasoning and agentic work.
Neon costs $3 for each million input tokens and $15 for each million output tokens, and cached inputs are discounted by 95 percent. At its high setting it ties the flagship Carbon (high) at 60 points. On the index it averaged $6.44 per task against $7.76 for Carbon, and took 21 minutes per task on average versus 36. On the Coding Agent Index, Neon (xhigh) led with 59 points, one point above Claude Opus 5.5 and Claude Code running Fable 5.2.
Developers can reach the model via Google AI Studio, the Gemini API and the Gemini Enterprise Agent Platform. Pro and Ultra subscribers can also use Neon inside Antigravity and the Gemini app. It offers a one-million-token context window and handles text, images, video and speech. Neon replaces Argon, and Gemini 3.1 Pro is being retired. Google itself supplied many of the top results, which outside testers have yet to verify.
Context
Google opened the Gemini 4 lineup at the end of September with Argon, first offered only to cybersecurity defenders in its Fairwind program. Carbon, the flagship aimed at the most difficult extended workloads, opened to every paying developer last week. Hydrogen, the fast successor to Gemini 3.8 Flash, now powers Google Search and scores 51 on Artificial Analysis. Antigravity uses an orchestrator that routes tasks across the models automatically.
El boletín
La semana tecnológica, en un solo correo.
Las crónicas que merecen su tiempo, cada semana. Cancele la suscripción cuando quiera.
