explainx.ai350K+ learn from us
TrendingAI News TodayExplainx for Business
Pricing
explainx.ai

Upskill in AI — 16 free pathways, live workshops & bootcamps, and 50+ courses from practitioners. Plus the skills, tools, and MCP servers to practice on.

follow us

follow on google

Add explainx.ai as a preferred source

corporate training

support@explainx.ai

get started

Find your pathTake Free Evaluation

community

Join the community

learn

mind: share how you thinkpathways — start freeworkshopsbootcampscoursescompare Explainxcertificationsmock testsexplainx universitycorporate traininglearn skills & mcp

discover

skillsmcp serversexplainx mcpestacktoolsmdx readeragentsllmsdesignsdictionarypeopleagi trackerfelony benchranks

company

aboutvisionmissionteaminstructorsteach on explainxpartnershipscommunityhackathonscareers

content

daily AI newsstate of AI — live resultsblogreleasespromptsgeneratorsresource libraryfor LLMsexplainx.ai kids

solutions

all solutionsdeveloper upskillingmarketing upskillingproduct manager upskillingleadership upskilling

newsletter · weekly

Get AI news, tools, and insights in your inbox.

supportcontactprivacytermsdata rightshow we create contentsubmission guidelines

© 2026 AISOLO Technologies Pvt Ltd

explainx.ai

On this page

  • Update (November 19, 2026): Gemini app tier and reported API prices
  • What is verified today
  • Why SKUs matter anyway
  • What press reports add (unverified primary)
  • How builders should prepare
  • What would confirm GA
  • Timeline of Neon sightings (secondary sources)
  • How to test without betting production
  • Router and Foundry implications
  • Competitive framing
  • What Google has not said
  • FAQ alignment with Search Console intent
  • Provisioned throughput and enterprise contracts
  • Multimodal SKU rows watchers reported
  • Naming confusion with Periodic Labs Neon
  • Communication template for executives
  • Historical pattern: SKU before blog
  • FinOps checklist when Neon goes live
  • Safety and cyber tiers
  • Developer experience on Antigravity CLI
  • When to update this post
  • Coordination with Gemini app consumer tiers
  • Related reading
← Back to blog

explainx / blog

Gemini 4 Neon on Vertex: What's Claimed vs What Google Docs Show

Google Gemini, Vertex AI, Pricing

Part of Google Gemini and DeepMind

Gemini 4 Neon appeared in Google Cloud SKUs at $3/$15 per million tokens on Nov 18, 2026. Vertex model pages and release notes do not list it yet. What is verified.

Nov 18, 2026·9 min read·Yash Thakker
add explainx.ai
go deep
Gemini 4 Neon on Vertex: What's Claimed vs What Google Docs Show

Gemini 4 Neon may be Google''s next mid-tier reasoning model, but November 18, 2026 evidence is mostly billing paperwork, not a clean launch post. Trackers reported gemini-4-neon SKUs in Google Cloud''s catalog at about $3 per million input tokens and $15 per million output tokens for text, while https://cloud.google.com/vertex-ai/generative-ai/docs/models/gemini/4-neon returned 404 when we fetched it this run.

This article separates what third parties claim, what SKUs suggest, and what Google''s documentation still omits.

Update (November 19, 2026): Gemini app tier and reported API prices

9to5Google reported on November 18, 2026 that Google is rolling out Gemini 4 Neon as a mid-stack model in the Gemini app, positioned between Hydrogen and Carbon, and that it replaces Argon in that consumer surface. The same summary cites $3 per million input tokens and $15 per million output tokens for API-style billing, aligning with the SKU leak this post tracked on Vertex paperwork.

What that does not automatically confirm:

  • A live Vertex generateContent call for gemini-4-neon in your project (the dedicated model page still 404'd on earlier checks).
  • GA language on Google's model versions table without a dated release note.

Builders should treat consumer-app availability and enterprise API GA as separate gates. Test your own project ID before you retune agents that depended on Argon pricing or latency.

New FAQ angle: If your search is "Gemini 4 Neon price," the $3/$15 figures are now corroborated by 9to5Google's November 18 report on the Gemini app rollout, but still trace to secondary reporting rather than a pricing table we could fetch on cloud.google.com during this update.

What is verified today

Model version staircase with newest step marked, for SKU leaks ahead of doc updatesModel version staircase with newest step marked, for SKU leaks ahead of doc updates

table · 2 cols
ClaimStatus (Nov 18, 2026)
SKU rows for gemini-4-neonReported by multiple model-tracking accounts; we did not independently download Google''s entire SKU export
Callable Vertex endpointNot verified—model doc URL 404
Listed on model versions docNo Neon row in the family table we read via search cache
Official Google blog postNone found
Gemini web app upsell stringReported Nov 16 then removed by some checks cited in trade press

Google''s documented Gemini 4 lineup still centers Hydrogen, Carbon, and Argon—same week Carbon topped industry indices at launch.

Why SKUs matter anyway

Google has historically published Cloud billing SKUs shortly before public model announcements. Leaked pricing preceded prior Gemini tiers. SKU rows are strong paperwork signals, not shipping confirmation.

Neon''s rumored $3/$15 text price would sit:

  • Above Hydrogen (~$1.50/$7.50 in comparison posts)
  • Below Carbon''s standard $4/$20 trajectory documented for Carbon
  • Near Anthropic Sonnet-class $2/$10 competitive band mentioned in trade coverage

Until token prices appear on Google''s official pricing tables, budget as unconfirmed.

What press reports add (unverified primary)

Some outlets published Vertex-style marketing copy describing Neon as "fast reasoning" with "Carbon-close" coding and agentic quality at Flash-tier speed. We treat that language as secondary until it lives on cloud.google.com without 404.

Antigravity CLI references to a gemini-pro-orchestrator-neon variant are developer-tool sightings, not GA proof.

How builders should prepare

Branching model checkpoints, for pinning IDs until Neon appears in official tablesBranching model checkpoints, for pinning IDs until Neon appears in official tables

  1. Pin model IDs in configs; do not alias neon until GA.
  2. Run shadow traffic on Hydrogen vs Carbon today; Neon will likely slot between them on cost/latency.
  3. Watch December Carbon price changes—Neon positioning is relative to that move.
  4. If you depend on Gemini Decisions API for routing, add Neon only after router docs list it.

What would confirm GA

We will treat Neon as shipped when all occur:

  • Non-404 model page under Vertex docs
  • model-versions table row with release date
  • Working generateContent for gemini-4-neon in a standard project
  • Pricing row on Google''s public Gemini API price sheet

Until then, headline "Neon GA" posts are claimed vs verified stories—this one included.

Timeline of Neon sightings (secondary sources)

Independent trackers described a rough sequence we have not fully re-verified primary-by-primary:

table · 3 cols
Date (2026)SightingSource type
Nov 16gemini-4-neon string in Gemini web upsell UICommunity checks cited in trade press
Nov 18 ~03:00 ETCloud SKU rows live for text modalitiesModel-tracking accounts
Nov 18Antigravity CLI orchestrator variant nameDeveloper tool release notes cited in press
Nov 18Some outlets claim Vertex model page textSecondary tech press, not fetched live here

Use this table to prioritize what to re-check hourly once you depend on Neon in production.

How to test without betting production

Run this sequence in a disposable GCP project:

  1. Enable Vertex AI API and billing alerts.
  2. Attempt generateContent with model: gemini-4-neon via Gen AI SDK.
  3. If you receive 404 model not found, stop—SKU alone is insufficient.
  4. If it succeeds, capture latency, price on invoice, and safety settings immediately; Google may change flags during soft launch.
  5. Compare quality on your coding and agent eval slices against Hydrogen and Carbon at fixed token caps.

Document results internally; do not treat press screenshots as API contracts.

Router and Foundry implications

Microsoft Ignite highlighted model routers that pick cheaper models per task. Google teams likely want a Neon-class slot between Hydrogen and Carbon for high-volume agent steps.

Until Neon is real in API listings, keep routers pinned to documented IDs only. Shadow traffic with claimed IDs is how surprise bills happen when soft launches flip on region by region.

Competitive framing

Trade articles position Neon against Claude Sonnet-tier $2/$10 pricing. Even if SKUs match at launch, total task cost still depends on tokens per task, thinking budgets, and grounding calls.

Run cost per successful task evals, not only per-million-token list prices, as we recommend in GPT-6.2 comparison posts.

What Google has not said

As of this run Google has not published, in sources we could fetch:

  • Neon safety tier or cyber allowlist posture
  • Context window and thinking defaults
  • Retirement policy for earlier Gemini 3.x routes
  • Whether Neon replaces rumored Krypton upgrade names

Absence is not denial; it is incomplete paperwork. Plan architecture for Hydrogen + Carbon until Neon is explicit.

FAQ alignment with Search Console intent

Teams searching "Gemini 4 Neon" need one sentence up front: SKUs appeared; Vertex GA is unconfirmed in Google's model index we checked. That is the honest SEO answer until Google edits official docs.

Provisioned throughput and enterprise contracts

Some press snippets mention Provisioned Throughput for Neon. Enterprise buyers should wait for orderable PT SKUs tied to a model ID that resolves in API, not press copy alone.

Until then, negotiate Carbon and Hydrogen PT commitments you can actually invoke.

Multimodal SKU rows watchers reported

Trackers claimed separate billing SKUs for image, video, and audio modalities with the Neon name. Multimodal pricing can diverge sharply from text-only rows.

When Google publishes tables, recompute agent pipelines that fan out to vision steps—our multimodal agent posts show how fast bills compound.

Naming confusion with Periodic Labs Neon

explainx.ai also covers Periodic Labs Neon science models—a different Neon brand in the same news cycle. Disambiguate in internal docs: Gemini 4 Neon (Google) versus Periodic Neon (science lab).

Communication template for executives

Use this one-liner until GA: "Neon is in Google's billing catalog per third-party trackers; our Vertex calls still 404 the model page—no production switch yet."

Historical pattern: SKU before blog

Google's recent Gemini launches often show billing SKUs, AI Studio leaks, then model cards. Carbon and Hydrogen followed that rhythm within days. Neon may still be mid-sequence.

Teams burned by early Krypton rumors should apply the same patience: paperwork is not press release.

FinOps checklist when Neon goes live

  1. Add per-model budget alerts in GCP billing.
  2. Split dev and prod projects so experiments cannot exhaust quotas.
  3. Log model ID per request in your proxy—even a one-day misconfiguration can choose the wrong tier.
  4. Revisit batch and flex SKUs Supercite-style trackers mention at half price tiers; confirm eligibility in official pricing.

Safety and cyber tiers

Carbon and Argon ship cyber-permissive variants behind allowlists. Neon may or may not mirror that split. Security teams should not assume one Neon ID covers red-team workflows until Google documents tiers.

Developer experience on Antigravity CLI

Press reports tie Antigravity orchestrator Neon strings to coding agent CLIs. That is a hint Neon targets agentic orchestration, not only chat widgets.

Watch Microsoft's parallel Foundry router story and route expensive steps deliberately.

When to update this post

We will add an Update section when any of these occur: live Vertex model page, release-notes entry, working API in two regions, or Google blog naming Neon GA.

Coordination with Gemini app consumer tiers

SKU leaks coincided with Gemini subscription upsell strings appearing and disappearing in mobile clients. Consumer and enterprise launches may be decoupled: Neon might hit Vertex first, Gemini app later, or the reverse.

Architect integrations against API model IDs, not app marketing names, until Google publishes a single source of truth.

Subscribe to Google Cloud release notes RSS and set an internal Slack webhook for strings gemini-4 so engineers learn Neon GA before finance learns from the invoice.

If your org already committed spend on Carbon introductory pricing, model Neon as a potential downgrade path for volume tiers while Carbon handles hardest eval slices—but only after side-by-side quality gates on your own tasks, not vendor marketing adjectives like fast reasoning. Revisit contracts when official pricing posts land.

Partners building on Vertex AI access patterns should snapshot current Hydrogen and Carbon quotas before Neon redirects traffic, so capacity planning stays grounded in models that already answer today. Export billing CSVs nightly while SKU names are in flux. Tag finance owners in the same ticket as engineering leads so invoice surprises do not land in silos. Keep a living checklist of which model IDs each microservice calls. Review it after every Google pricing email lands internally as well.

Related reading

  • Gemini 4 Carbon Vertex access
  • Gemini 4 Hydrogen internal tests
  • Hydrogen vs Argon vs Carbon comparison
  • Google AI Mode temporary threads same week
  • GPT-6.2 competitive context

Sources: Google Cloud model versions documentation (fetched November 18, 2026); attempted fetch of https://cloud.google.com/vertex-ai/generative-ai/docs/models/gemini/4-neon returned 404; SKU pricing attributed to independent trackers including Testing Catalog's November 18 report and OfficeChai's SKU breakdown, not verified against a live Google price sheet in this run.

Spotted something out of date? Let us know.
Daily briefFrom Yash · 350K+ students taught

Catch up on AI

One AI story explained in plain English, plus one thing to try. Every day, free. Unsubscribe anytime.

  • ✓ Free
  • ✓ One email a day, no spam
  • ✓ Unsubscribe anytime
Yash Thakker

Written by

Yash Thakker

Yash is an AI expert with over 300K learners. Join his workshops →

View Yash Thakker in People in AI →

Related posts

Nov 18, 2026

GAO Covert Test: ChatGPT, Claude, Gemini and Grok Planned Viable Attacks in 12 Hours

On November 18, 2026 the Government Accountability Office released a performance audit that used covert prompts to test ChatGPT, Claude, Gemini and Grok. Investigators report viable improvised-explosive or electrical-grid attack plans within 12 hours after bypassing initial refusals, with bomb-squad review on the IED side.

Nov 18, 2026

Gemini App on Android Go: Google Brings Live and Images to Low-RAM Phones

On November 18, 2026 Google said it re-architected the Gemini app so core features run on lower-memory Android Go edition hardware. Users get Gemini Live in more than 45 languages, camera-based help on notes and forms, and Nano Banana image creation on supported phones.

Nov 17, 2026

Gemini Temporary Chats Now Kept ~2 Months: Privacy Hub Change Explained

Google's November 2026 Gemini Privacy Hub raises retention for temporary chats and History-off conversations from about 72 hours to around two months, while splitting History from AI training controls.