Particle.news

Google Launches Gemini 4 Neon as Mid‑Tier Model for Faster, Cheaper Complex Workflows

The model promises near‑Carbon reasoning at Flash speeds to lower per‑task cost for high‑frequency coding, research and enterprise tasks.

Overview

  • Google made Gemini 4 Neon available on Wednesday, November 18, 2026, through the Gemini API, Google AI Studio, Antigravity and the Gemini app for AI Pro and AI Ultra subscribers.
  • Neon is priced at $3 per million input tokens and $15 per million output tokens with cached input discounted to $0.15 and carries a 1,000,000‑token context window to handle long multi‑file workflows.
  • Google positions Neon as a mid‑tier orchestrator that can direct Hydrogen subagents in Antigravity and trade off four 'thinking' levels so teams can balance reasoning depth against token cost.
  • Company and third‑party tests cited in the launch show strong benchmark results, including a high placement on Arena.ai’s Agent Arena Index, but broader independent verification of Google’s claims is still limited.
  • Google will gate a specialized 'Neon for cyber' build to vetted defenders through its Fairwind Program and a December 1 pricing change for Argon and Carbon will widen Neon’s price advantage for production workloads.