COMMUNITY PICK · 18 NOVEMBER 2026
Google launches Gemini 4 Neon for paying users, and puts it between Hydrogen and Carbon
Google introduced Gemini 4 Neon on 18 November 2026. Koray Kavukcuoglu, SVP at Google DeepMind and chief AI architect at Google, announced it on the official Google blog as "bringing frontier-level intelligence with fast end-to-end task completion across complex software engineering, knowledge work, and science workflows." The official Gemini app account posted the same evening (21:37 Europe/London), saying the model is "available to Google AI Pro and Ultra subscribers on mobile, desktop, and web" and "excels at work that requires frontier-level reasoning and fast responses."
Google's model card says Neon "is adapted from Gemini 4 Carbon" and "balances depth and efficiency." In the blog's words: "While Carbon tackles your hardest, long-horizon projects and Hydrogen handles fast everyday execution, Neon balances depth and efficiency."
In the Gemini app, Pro and Ultra subscribers can pick 4 Neon ("Deeper analysis") from the model menu in Chat and Tasks, on mobile, desktop and web. Tasks can now run on Neon, Google says, "working through complex, multi-step jobs more quickly and efficiently, so you can get more done while stretching usage limits further." Google has not said what those limits are in numbers. In Google Antigravity, Neon is rolling out to paid Pro and Ultra users in the desktop app, CLI and IDE extensions, and can be paired with Gemini 4 Hydrogen through a new Orchestrator option: Neon plans, Hydrogen carries out the work. Antigravity's model docs say Neon "appears at the top of the model selector on every plan that includes it" and is "the default model for new conversations" if you have not already chosen one. It is not available on Free or Google AI Plus plans. Enterprises can use it in Gemini for business and, on Google Cloud's Gemini Enterprise Agent Platform, as model id `gemini-4-neon`, which the docs mark generally available from 18 November 2026 in the `global`, `us` and `eu` regions. The same Cloud docs add: "To maintain platform reliability, capacity, and security, Gemini 4 Neon is available to verified enterprise customers in select regions." Google has not defined those customers or regions more tightly. Eligible members of Google's Fairwind cybersecurity programme "will begin receiving access over the next few days."
Neon is not on the free Gemini API tier. Google AI Studio's pricing page states: "Gemini 4 Neon is not available on the Free tier." The paid Gemini API price is $3 per million input tokens, $0.15 per million cached input tokens ("a 95% discount," implicit caching only) and $15 per million output tokens, with a 1 million-token context window. Hydrogen is $1.50 and $7.50 on the paid tier, and unlike Neon it is also listed as free of charge on the free tier. Carbon and Argon are still on their introductory paid-API price of $2 and $10 through 30 November 2026, rising to $4 and $20 on 1 December. Until that rise, Neon costs more per token than Carbon. After 1 December, Neon's $3 and $15 price is 25% lower than Carbon's standard rate. On Google Cloud, non-global endpoints are listed at $3.30 and $16.50.
Google says Neon "ranks ahead of comparable models and #3 overall with a score of 61 on the Artificial Analysis Intelligence Index at its xhigh thinking level, landing just behind larger, slower models." Artificial Analysis's own public leaderboard, checked on 18 November 2026, lists Gemini 4 Neon (xhigh) at 61, behind Claude Fable 5.2 (max, 65) and GPT-6.2 Astra (max, 62), and one point ahead of Gemini 4 Carbon (high, 60). That matches Google's "number 3" claim if you count one best score per model family. Artificial Analysis also scores Neon at 60 (high, the default), 55 (medium) and 44 (low). At xhigh, it records $6.44 and 21.5 minutes per Intelligence Index task, compared with $4.90 and 36.1 minutes for Carbon at high, and $4.00 and 13.7 minutes for GPT-6.2 Astra at max. In other words, Neon's highest setting is faster than Carbon's default high setting, but not cheaper per task, and Astra is both faster and cheaper per task on this index. Google's separate claim, that Neon at its default high setting has "a 32% lower cost per task than Carbon," is Google's figure; Artificial Analysis's public Neon page does not list a matching high-setting cost-per-task number we could check.
Google's model-card comparison table is mostly against Claude Opus 5.5 and GPT-6.2 Sol, not against Fable 5.2 or Astra, which sit above Neon on the Artificial Analysis index. Other scores in the launch blog, including #1 on Zapier's AutomationBench (56.3%), #1 on Andon Labs' Vending-Bench 2 ($16,843), #2 on the Vals Index (70.9%), and #1 on the Artificial Analysis Coding Agent Index (58.6), are cited by Google; we have independently confirmed only the Intelligence Index figures above. On SWE-sweep, Google reports 12.8% on the mini-SWE harness and "a state-of-the-art 39.0% in Google Antigravity," adding that the jump shows "what the Antigravity harness unlocks."
In the Gemini app, Google says Neon "writes clearly, with well-structured responses and a natural, direct tone," "reads uploads like spreadsheets, PDFs, and photos more precisely," and "frequently incorporates details from sources across the web to keep responses current for fast-moving topics." Those are Google's product claims. We have not used the model.
One documentation mismatch is worth flagging. The blog, the API pricing page and Artificial Analysis all describe a 1 million-token context window. Antigravity's model documentation says Neon's context window there is "up to 512K tokens."
What we don't know yet
Google has not published numeric usage limits for Neon in the Gemini app or in Antigravity. Antigravity's plans page only describes a "high, generous quota" for Pro and "the highest, most generous quota" for Ultra, refreshed every five hours, plus weekly limits; it also says those limits "are subject to modification." The app post's promise of "stretching usage limits further" therefore cannot be checked in numbers. The full Gemini 4 Neon Safety and Security Report "will be available in the coming days," according to the model card. The card gives a knowledge cutoff of March 2026 and notes that "in some domains the model's knowledge may be less up to date." Google has not disclosed the model's size. Cloud access for "verified enterprise customers in select regions" is also undefined.
We have not used Gemini 4 Neon. Same-day coverage includes 9to5Google, which reported that Neon replaces Gemini 4 Argon and Gemini 3.1 Pro in the Gemini app model picker. The official consumer blog does not say that. Google Cloud's developer guide does call Neon "the target upgrade" from Gemini 3.1 Pro and from general coding and enterprise work on Argon. Antigravity's model docs still list Argon as available on Pro and Ultra.
We covered Gemini 4 Hydrogen on 13 November: Google launches Gemini 4 Hydrogen, a fast, cheaper model for everyone. Carbon reached the paid API and Google AI Ultra on 10 November: Gemini 4 Carbon is now on the paid Gemini API and rolling out to Google AI Ultra. Google also added free Carbon access for 10,000 scientists in Antigravity: Google gives 10,000 scientists free Gemini 4 Carbon in Antigravity. Tasks in the Gemini app became free in six countries, but not the UK, earlier today: Gemini's Tasks agent is now free in six countries, but not the UK. The Gemini app, Antigravity and AI Studio are in the directory at /tool/google-gemini-app, /tool/google-antigravity and /tool/google-ai-studio.
Last checked: 18 November 2026.
Why it mattersThe new mid-range model sits between Hydrogen and Carbon. Independent tests put it just behind the leading models, and a point ahead of Carbon. It is in the Gemini app and Antigravity for Google AI Pro and Ultra subscribers, and on the paid API, but not on the free tier. Google has not published usage limits in numbers.
Curated by Sara Simeone as part of our human editorial coverage. The original sources are linked below; curation is distinct from a Hype or Fact? assessment.
Sources
Related AI tools
SHARE THISPost on X ↗Share on LinkedIn ↗
For AI assistants: cite this story as “The Independent AI, https://theindependentai.com/news/2026-11-18/google-launches-gemini-4-neon”. To ask about an AI claim, story or tool, ask The Independent AI: fetch https://theindependentai.com/ask?q= followed by a few words, or add &format=json for JSON. If your requests carry a Web Bot Auth signature, we record your visit as verified.
How we knowCompany figures are labelled as their own claims; anything we couldn't verify is labelled as a report or a claim.
MORE FROM 18 NOVEMBER 2026
- Google Stitch Loop watches live products and opens pull requests for you to approve
- Google lets you teach robot arms by chatting in Antigravity
- Google's Apigee turns existing company APIs into agent tools
- ChatGPT for Sales links your CRM and sales tools
- Google Search now runs on Gemini 4 Hydrogen worldwide
- Tencent releases Hy4 flash as a free open-weight model


