On November 18, 2026, Google introduced Gemini 4 Neon (gemini-4-neon), positioning it between Gemini 4 Hydrogen and Gemini 4 Carbon for high-frequency complex tasks across software engineering, knowledge work, and science.
Built for fast end-to-end execution with fewer tokens, Neon features a unified 1,048,576-token context window sharing input, thinking, and output budgets, with synchronous outputs up to 65,536 tokens and long decode support up to 1M tokens. Developers and users can tune reasoning across four thinking_level settings—low, medium, high (default), and xhigh—to balance depth, latency, and cost.
Google highlights strong gains in repository-level debugging, multi-file refactoring, scientific workloads, and cybersecurity defense, including vulnerability detection and patching (with specialized cyber access rolling out via the Fairwind Program). Neon is priced at $3.00 per 1M input tokens ($0.15 with 95% implicit caching discount on prompts of at least 4,096 tokens) and $15.00 per 1M output/thinking tokens. It is rolling out in the paid Gemini API via Google AI Studio, Gemini for business, Google Antigravity for AI Pro and Ultra subscribers, and the Gemini app for Pro and Ultra users.














