Home / Industries in Trend / Ai vr and automation / Gemini 4 Neon Brings New Co...

visionaries Network Team

20 November, 2026

ai vr and automation

Gemini 4 Neon, a new AI coding agent from Google DeepMind, targets complex coding tasks, software development, and long-running agent workflows

Today, Google DeepMind unveiled Gemini 4 Neon, its latest AI coding agent, aimed at tackling complex software development, research and long coding sessions. It went live on Nov. 18, 2026, and the model is designed to follow multi-step instructions, use tools and keep track of progress on more lengthy tasks.

Gemini 4 Neon’s Artificial Analysis Intelligence Index score for Gemini 4 Neon at its xhigh thinking level was 61, placing it third overall. It scored 59 on the Coding Agent Index, which is three points higher than the Gemini 4 Carbon. It can do software engineering, debugging, changing code, and anything that requires sustained reasoning, not just a single prompt. Google's Gemini 4 Neon announcement.

Handling Longer Coding Sessions

Gemini 4 Neon is made for coding tasks that involve multiple steps and tools. It can track progress, avoid unnecessary file reading, and correct as you go.

These capabilities are useful for AI agent workflows, where a model needs to plan a task, perform it, review the results and move to the next step. Google says Neon can share progress and keep track of context over long conversations, which could help developers work on larger projects.

Working With Google Antigravity AI

The model can be used standalone, or as an orchestrator in Google Antigravity, where multiple Hydrogen subagents can be working on different parts of a task in parallel. Instead of doing every single step one by one, developers are able to break work down between agents.

Developers are able to use agent-based tools to code, edit files, debug, and collaborate on software projects with Google Antigravity AI. Gemini 4 Neon can also react to changes in instructions while a task is underway. Google Antigravity.

Pricing and Availability

Gemini 4 Neon costs $3/million input tokens and $15/million output tokens. We charge $0.15 per 1M cached input tokens. It has a context window of up to one million tokens, which means it can take in large amounts of code and documentation.

The model is available through the Gemini API, Google AI Studio, and Google Antigravity, with access depending on the product and subscription. Official Gemini model details.

 FAQs

1. What is Gemini 4 Neon?

Gemini 4 Neon is a Google DeepMind model designed for coding, technical research, complex reasoning, and multi-step tasks.

2. What can Gemini 4 Neon do for coding?

It can assist with code editing, debugging, software engineering, and tasks involving multiple files and development steps.

3. What is Google Antigravity AI?

Google Antigravity is an agent-based development platform where AI can assist with coding tasks, work with tools, and handle different stages of software development.

4. What are AI agent workflows?

AI agent workflows allow a model to complete tasks in stages, including planning, using tools, checking results, and making changes without requiring a new instruction for every step.

5. What is the API pricing for Gemini 4 Neon?

The API costs $3 per million input tokens and $15 per million output tokens. Cached input tokens cost $0.15 per million.