AI Watch

Gemini 4 Neon Launches Across Antigravity, Gemini App, and API: Plans, Thinking Levels, and Pricing

By Lesia Kapliuk · Published

TL;DR: Google released Gemini 4 Neon (gemini-4-neon) on November 18, 2026. It is the new default model in Antigravity on Google AI Pro and Ultra (512K context), adds Orchestrator (Neon), and replaces Argon under "Deeper analysis" in the Gemini app. In the Gemini API it is paid-only at $3/$15 per 1M tokens, with a 1M unified context window and low/medium/high/xhigh thinking.


What is Gemini 4 Neon?

According to Google's What's new in Gemini 4 Neon guide and Gemini 4 Neon model page, gemini-4-neon is built for "frontier-level intelligence with fast end-to-end task completion across complex software engineering, knowledge work, and science workflows." Google positions it between Gemini 4 Hydrogen and Gemini 4 Carbon for high-frequency complex tasks.

Unlike gemini-4-hydrogen, which launched on the Gemini API free tier (with a small daily cap), gemini-4-neon is available only on the paid tier of the Gemini API (generateContent and the Interactions API) and for Google AI Pro and Ultra users in Google AI Studio. The release also lands on Google Antigravity's first anniversary, one year after Antigravity launched on November 18, 2025.


Gemini 4 Neon in Google Antigravity: plans, thinking levels, Orchestrator

Google's Antigravity models page lists Gemini 4 Neon and Orchestrator (Neon) across the Antigravity App, CLI, and IDE:

ModelFree & Google AI PlusGoogle AI ProGoogle AI UltraEnterprise
Gemini 4 Neonnoyes (paid or eligible trial)yesyes
Orchestrator (Neon)noyes (paid or eligible trial)yesno
Gemini 4 Carbonnonoyesyes
Orchestrator (Carbon)nonoyesno
Gemini 4 Argonnoyes (paid or eligible trial)yesyes
Orchestrator (Argon)noyes (paid or eligible trial)yesno
Gemini 4 Hydrogennoyes (paid, non-trial)yesyes
Gemini 3.8 Flashyesyesyesyes

Plan table from Google's Antigravity models page, read November 18, 2026.

A few details from the documentation and what we verified on our own paid Google AI Pro account in Antigravity CLI 1.5.11:

  • New default model: Antigravity places Gemini 4 Neon at the top of the model selector and uses it as the default model for new conversations if you have not already picked another model.
  • Available on eligible trials: Google's table marks Gemini 4 Neon (like Argon) as available on trials with eligible premium devices, whereas standalone Gemini 4 Hydrogen is restricted to non-trial paid Pro subscriptions. Notably, the same table also marks Orchestrator (Neon) as available on eligible trials even though Orchestrator hands execution steps to Hydrogen; Google's page does not explain how trials handle the Hydrogen step.
  • Four thinking levels: In the Antigravity app and IDE, Neon offers Low, Medium, High (default), and Extra. In Antigravity CLI (/model and agy models), those same four levels appear as low, medium, high, and xhigh:

Antigravity CLI 1.5.11 model picker showing Gemini 4 Neon with Effort levels low, medium, high and xhigh

> agy models
gemini-4-neon-high            Gemini 4 Neon (High)
gemini-4-neon-xhigh           Gemini 4 Neon (XHigh)
gemini-4-neon-medium          Gemini 4 Neon (Medium)
gemini-4-neon-low             Gemini 4 Neon (Low)
gemini-pro-orchestrator-neon  Orchestrator (Neon)
  • Argon is still in Antigravity: On our Pro account, agy models still lists all three gemini-4-argon levels and gemini-pro-orchestrator-g1-pro (Orchestrator (Argon)) alongside Neon.
  • 512K context window in Antigravity: While the Gemini API gives gemini-4-neon a 1M-token context window, the Antigravity models page lists up to 512K tokens inside Antigravity.
  • Orchestrator (Neon): Like Orchestrator (Argon) and Orchestrator (Carbon), Orchestrator (Neon) plans and reviews with Neon while handing implementation work to Gemini 4 Hydrogen. Starting in Antigravity CLI 1.5.10, the interactive /model picker groups all Orchestrator variants under a single Orchestrator row with a Planner choice, while agy models exposes gemini-pro-orchestrator-neon directly.

(See What we haven't tested yet for quota-pool and benchmark gaps.)


Gemini app: Neon replaces Argon

On our Google AI Pro account in the Gemini web app (gemini.google.com/app) on November 18, 4 Neon replaced 4 Argon in the model selector:

  • Auto — Adapts to your needs
  • Everyday help — 4 Hydrogen
  • Deeper analysis — 4 Neon

We observed this change on our own account; model rollouts in the Gemini app can be gradual across regions and accounts. Unlike Antigravity and the API (which offer four levels including Extra / xhigh), the Gemini web app's Thinking level submenu for 4 Neon shows three options: Low (Fast and efficient), Medium (Balanced depth), and High (Extra thorough — not to be confused with Antigravity's fourth Extra level).

Gemini web app model selector on Google AI Pro showing 4 Neon under Deeper analysis with Low, Medium and High thinking levels


Gemini 4 Neon API pricing and specs

In our check of the Gemini API model list and Google AI Studio's documentation on November 18, gemini-4-neon is live with the following specifications:

Google AI Studio gallery banner "Gemini 4 Neon is here" and "Build with Gemini 4 Neon" example apps

Gemini 4 Neon (gemini-4-neon)
Context window1,048,576 tokens (unified input + output)
Max output65,536 tokens (default synchronous); up to 1,048,576 total with long decode (shared with input)
Inputs / OutputText, Image, Video, Audio, PDF / Text
Thinking levelslow, medium, high (default), xhigh
Free tierNot available
Paid tier (Standard)$3.00 / 1M input tokens, $15.00 / 1M output tokens (thinking included)
Batch & Flex$1.50 / 1M input, $7.50 / 1M output
Priority$5.40 / 1M input, $27.00 / 1M output
Context cachingImplicit only ($0.15 / 1M standard; no explicit caching)

Source: AI Studio pricing, model page, and What's new in Gemini 4 Neon, read November 18, 2026.

One temporary pricing quirk to watch through November 30: Neon's $3.00 / $15.00 standard rate is 2× Gemini 4 Hydrogen ($1.50 / $7.50) and 75% of Gemini 4 Carbon's regular rate ($4.00 / $20.00). However, both Gemini 4 Argon and Gemini 4 Carbon are on a $2.00 / $10.00 introductory rate through November 30, 2026 (moving to $4.00 / $20.00 on December 1), so Argon and Carbon are temporarily cheaper per token in the API than Neon until December 1. Batch and Flex halve Neon's rate to $1.50 / $7.50; see Gemini API pricing tiers explained.

In the API, gemini-4-neon supports generateContent, batchGenerateContent, countTokens, and the Interactions API, along with code execution, computer use, function calling, structured outputs, file search, and grounding with Google Search and Google Maps. Live API, explicit context caching, image generation, and audio generation are not supported. In AI Studio's Build mode, Google also switched the built-in image and video understanding feature templates from gemini-4-argon to gemini-4-neon on November 18.


Gemini 4 Neon vs Hydrogen vs Argon vs Carbon

FeatureGemini 4 HydrogenGemini 4 NeonGemini 4 ArgonGemini 4 Carbon
API paid price (in / out per 1M)$1.50 / $7.50$3.00 / $15.00$2.00 / $10.00 through Nov 30 ($4 / $20 after)$2.00 / $10.00 through Nov 30 ($4 / $20 after)
API free tierYes (low daily cap)NoNoNo
Context window (API)1M (unified)1M (unified)1M (unified)1M (unified)
Context window (Antigravity)—Up to 512K——
Thinking levels (API / Antigravity)low, medium, high, xhigh (Extra)low, medium, high, xhigh (Extra)low, medium, highlow, medium, high
Antigravity plansPro (non-trial), Ultra, EnterprisePro (paid or eligible trial), Ultra, EnterprisePro (paid or eligible trial), Ultra, EnterpriseUltra, Enterprise

API prices and Antigravity availability as documented on November 18, 2026.


Gemini 4 Neon by plan: Free, Plus, Pro, Ultra, and API

  • Google AI Pro in Antigravity: You get standalone Neon (Low through Extra) and Orchestrator (Neon) alongside Hydrogen and Argon. New conversations default to Neon unless you have already chosen another model.
  • Google AI Pro in the Gemini web app: "Deeper analysis" switched from 4 Argon to 4 Neon on our account, with Low, Medium, and High thinking levels.
  • Free or Google AI Plus plans: Antigravity's models page marks Neon, Hydrogen, and Argon as unavailable on Free and Plus, leaving Gemini 3.8 Flash as the free Gemini option there.
  • Gemini API developers: There is no free tier for gemini-4-neon. Before December 1, check whether Argon or Carbon's $2/$10 introductory rate—or Hydrogen's $1.50/$7.50 rate—fits your task better than Neon at $3/$15.

How to select Gemini 4 Neon in Antigravity, the Gemini app, and AI Studio

  • Antigravity CLI: Run agy models to list the gemini-4-neon-* slugs and gemini-pro-orchestrator-neon, or open /model inside agy and use the Effort slider (low · medium · high · xhigh).
  • Antigravity App & IDE: Open the model selector under the prompt box—Neon sits at the top on eligible plans, with Low, Medium, High, and Extra.
  • Gemini web app: Open the model menu next to the prompt box and choose Deeper analysis (4 Neon).
  • Google AI Studio: Open New chat with Gemini 4 Neon (paid API tier or Google AI Pro/Ultra required).

FAQ

Is Gemini 4 Neon free?

No in the Gemini API and Antigravity Free/Plus tiers. gemini-4-neon is paid-only in the Gemini API, and Antigravity's models page excludes Free and Google AI Plus plans; you need at least Google AI Pro (or an eligible trial) in Antigravity and AI Studio.

How much does Gemini 4 Neon cost?

In the Gemini API Standard tier, gemini-4-neon costs $3.00 per 1M input tokens and $15.00 per 1M output tokens (including thinking tokens). Batch and Flex halve this to $1.50 / $7.50, Priority is $5.40 / $27.00, and implicit context caching is $0.15 / 1M.

What is the Gemini 4 Neon context window?

In the Gemini API, gemini-4-neon has a 1,048,576-token (1M) unified context window shared across input and output, with 65,536 default synchronous max output (up to 1M total with long decode). Inside Antigravity, Google documents a context window of up to 512K tokens.

Neon vs Hydrogen: which should I use?

Use Hydrogen for routine coding, fast iteration, and lower API cost ($1.50/$7.50 vs $3/$15), or Neon when a task needs deeper reasoning without stepping all the way up to Carbon. In Antigravity on Pro or Ultra, Orchestrator (Neon) lets Neon plan and review while Hydrogen handles implementation.


What we haven't tested yet

  • Which Antigravity quota pool Gemini 4 Neon and Orchestrator (Neon) draw from (Argon's weekly pool, Flash/Hydrogen's pool, or a separate pool).
  • Real-world coding speed, thinking token usage, and quality of Neon compared with Hydrogen, Argon, and Carbon.

Sources: agy models on Antigravity CLI 1.5.11 and our Google AI Pro account in the Gemini web app on November 18, 2026; our Gemini API model list check and AI Studio ListModels capture on November 18, 2026; Google's Antigravity models page, AI Studio changelog, Gemini 4 Neon model page, What's new in Gemini 4 Neon, and pricing, all read November 18, 2026. In the CLI screenshot we covered the account email; in the Gemini web screenshot we removed the browser extensions and account avatar. How we work: editorial policy.