Free to read. Sign up to save tools and get alerts when they change. Plus 1,000+ more AI tool profiles.

Sign up free
5 min read·Updated November 19, 2026

Gemini 4 Neon is the fourth model in Google DeepMind's Gemini 4 line, released on November 18, 2026 for fast, long-running coding, knowledge and science work. Google says it does a typical task for about a third less than Gemini 4 Carbon because it uses fewer tokens, though its 3 dollar and 15 dollar per million token list price is above Carbon's introductory rate through November 30. Unlike Carbon, it reaches Google AI Pro subscribers.

Share

Listen to this overview

Free preview · first 0:30
0:00 / 0:30

Unlock audio and more

Audio streaming, downloadable PDFs and certificates come with Plus and Pro.

Learning Objectives

  • Understand where Gemini 4 Neon sits in Google's Gemini 4 line, next to Argon, Carbon and Hydrogen
  • Know who can use Neon and why "cheaper than Carbon" is true per task but not per token
  • Read Google's benchmark and safety claims for what they are: reported by Google, with no outside test yet

What Is Gemini 4 Neon?

Gemini 4 Neon is the fourth model in Google DeepMind's Gemini 4 generation, released on November 18, 2026, after Gemini 4 Argon, Gemini 4 Carbon and the fast, low-cost Gemini 4 Hydrogen. Google describes it as frontier-level intelligence with fast end-to-end task completion, aimed at software engineering, everyday knowledge work and science workflows, with a 1 million token context window.

The pitch is efficiency rather than peak capability. Google says Neon ranks third overall on the Artificial Analysis Intelligence Index, with a score of 61 at its highest thinking setting, just behind larger and slower models, and that it reaches that level while spending fewer tokens per task than Carbon. In agent teams Google positions it as a fast orchestrator that directs parallel Hydrogen agents in Antigravity.

⚠️Warning

Cheaper per task, not per token, until December. Neon lists at 3 dollars per million input tokens and 15 dollars per million output. Gemini 4 Carbon's introductory API rate of 2 dollars and 10 dollars runs through November 30, 2026, so until then Carbon costs less per token. Google's claim is that Neon finishes a typical task for about 32 percent less than Carbon at its default setting, because it needs fewer tokens. Check that on your own workload before switching.

Benchmarks and Early Results

Every figure below is reported by Google, including results on other companies' benchmarks. No independent evaluation had been published at launch.

BenchmarkWhat it measuresGemini 4 Neon
Artificial Analysis Intelligence IndexOverall capability across tests61 (third overall, highest thinking setting)
Artificial Analysis Coding Agent IndexAgentic coding58.6 (first overall)
Vals IndexProfessional tasks70.9% (second)
Terminal-Bench 4.1Command-line agent tasks65.2%
AutomationBench (Zapier)End-to-end business tasks56.3% (first)
SEC-bench ProFinding software vulnerabilities90.5% (cyber version)

Google also reports first place on the Legal Research Bench and Tax Agent Bench, second place on Arena's new Agent Arena Index for steerability, and its lowest fabrication rate yet in internal Antigravity testing. On a test of real repository fixes it says Neon's score rises from 12.8 percent in a plain agent harness to 39 percent inside Antigravity, which says as much about the harness as about the model.

Safety and Security

Google says Neon keeps the protections it built for Carbon against cyber and biological misuse while declining fewer legitimate tasks, refusing about 2 percent of tasks on one terminal benchmark. It reports a 99.2 percent pass rate on its agent honesty test, shortcuts on 2.45 percent of tasks on its coding cheating test, the lowest of any Gemini 4 model, and no successful attacks on Gray Swan's indirect prompt injection benchmark at the highest thinking setting with its real-time safeguards on. A separate cyber version is rolling out to vetted defenders in the Fairwind Program.

Pricing & Access

Gemini API$3 input / $15 output per 1M tokens
  • Cached input 15 cents per 1M
  • 1M token context window
  • Also in Gemini for business
Google AI Pro$19.99/month
  • Neon in the Gemini app
  • Neon in Antigravity
  • Pro does not get Carbon
Google AI UltraSubscription
  • Neon and Carbon in the Gemini app
  • Neon and Carbon in Antigravity
  • Swarm agent teams in Antigravity
Fairwind ProgramBy application
  • Vetted cyber defenders
  • Separate cyber version
  • Rolling out after launch

Neon has four thinking levels so you can trade intelligence against token spend. Free users of the Gemini app do not get it.

Strengths

  • Strong results for the token spend: Google reports third place on the Artificial Analysis Intelligence Index and first on its coding agent index, at lower cost per task than Carbon
  • Reaches Pro subscribers: the strongest Gemini 4 model a Google AI Pro plan includes, since Carbon is Ultra-only
  • Built for long agent runs: Google says it cuts redundant file reads, reports progress as it works and handles long sessions reliably
  • Security and honesty results: low cheating and fabrication rates in Google's own tests, and a separate cyber version for defenders

Limitations & Considerations

  • All numbers are Google's: no outside benchmark had been published at launch, unlike Carbon, which SemiAnalysis's GDPx rated in November
  • Not the top model overall: by Google's own count it ranks third on the Artificial Analysis Intelligence Index, behind larger models
  • Per-token price is higher than Carbon's for now: through November 30 Carbon's introductory rate is lower, so the saving depends on Neon using fewer tokens on your tasks
  • Harness-dependent coding scores: some of its best coding results are measured inside Google's own Antigravity environment
  • No free tier in the app: free Gemini users do not get Neon
  • Gemini 4 Carbon — Google's strongest model, for the hardest coding, research and security work
  • Gemini 4 Argon — the first Gemini 4 model
  • Gemini Flash — Gemini 4 Hydrogen, the fast, low-cost model Neon can direct as subagents
  • Gemini — Google's consumer app, where Neon reaches Pro and Ultra subscribers
  • Google Antigravity — Google's agent-first coding environment

Getting Started

  1. Developers: select Gemini 4 Neon on the paid API tier in Google AI Studio, or in Gemini for business
  2. Consumers: with Google AI Pro or Ultra, pick Neon in the Gemini app's model menu
  3. Start at the default thinking level and raise it only for the hardest tasks
  4. Compare cost per finished task against Carbon on your own work before switching, especially before November 30

Key Takeaways

  • Gemini 4 Neon is the fourth Gemini 4 model, released on November 18, 2026 for fast, long-running coding, knowledge and science work
  • Google says it does a typical task for about a third less than Gemini 4 Carbon, because it uses fewer tokens
  • Its list price of 3 dollars and 15 dollars per million tokens is higher than Carbon's introductory rate until November 30
  • Google AI Pro subscribers get Neon, though not Carbon, and free users get neither
  • Every benchmark at launch was reported by Google, with no independent test yet

Keep track of the tools you’re evaluating

  • The AI Hub on a phone: a 12-day AI Skill Streak and an expanded Content updates alert listing the saved items that changed.
  • Recommended for you on a phone: nine personalised suggestions labelled Trending in AI news, On your saved list, and Popular.
  • My AI Tools on a phone: saved tools including GitHub Copilot and OpenAI Codex, each with an Updated badge.

Swipe for Recommended for you and My AI Tools

Your AI Hub — sample data.

📰Gemini 4 Neon in the News

Showing the only story where Gemini 4 Neon is tagged in Top AI Stories.

Other tools in Foundation Models & Open Source (12 of 85)

Show 7 more →

Other tools from Google

Show 22 more →
🧭Recommended for you

Optional detours — these connect to what you just read, and your next lesson will be waiting.