Google launched Neon, the fourth model in the Gemini 4 series, in just seven weeks, positioning it at a key point in its product matrix. The model features million-token context and output limits, four adjustable levels of thinking depth, and ranks third globally in comprehensive reasoning capability, leading in multiple programming, enterprise-level, and scientific research benchmarks. Its output speed has doubled, and the cost per task is roughly half that of competitors' flagship models, giving it outstanding cost-effectiveness and earning praise from developers and everyday users. Neon supports multi-agent collaborative scheduling, can autonomously complete complex engineering tasks, and can serve as an orchestration core to coordinate with the lightweight model Hydrogen for division of labor and collaboration, significantly reducing developer operating costs. Previously, Google AI faced scrutiny over product delays and lackluster performance. Recently, through organizational adjustments, the establishment of a high-frequency iteration pipeline, strengthened targeted training, and its self-developed seventh-generation TPU computing power foundation, it has quickly turned the situation around and returned to the core competitive ranks of the world's top-tier AI vendors.
