Skip to content
→ All articles

AI

Google Announces Gemini 4 Neon, Its New Model for Long Coding and Complex Agent Tasks

Header image from Google's official blog post introducing Gemini 4 Neon
Photo: Google

Google announced Gemini 4 Neon on November 18, 2026, a new Gemini 4 model built for long-running coding tasks, agentic work, and complex enterprise and scientific knowledge work. The announcement on Google's official blog was written by Koray Kavukcuoglu, SVP of Google DeepMind and Chief AI Architect at Google, who described the model as delivering frontier-level intelligence with fast end-to-end task completion.

Neon runs on a context window of up to 1 million tokens, letting it process spreadsheets, PDFs, charts, audio, and video in a single request. Google says the model ranks first on the Artificial Analysis Coding Agent Index with a score of 58.6%, and takes the top spot on both the Legal Research Bench and Tax Agent Bench, with strong results on finance, accounting, and spreadsheet-modeling benchmarks. Pricing is $3 per million input tokens (with a 95% discount bringing cached input tokens to $0.15) and $15 per million output tokens.

Google began rolling out Gemini 4 Neon the same day, through the Gemini API, Google AI Studio, Gemini for business, Pro and Ultra subscribers in the Gemini app, and Google Antigravity, its environment for coding agents.

Source: Google

Share WhatsApp X Telegram

Need filming at the level you just read about? I reply within 24 hours.

Start your project