Gemini 4 Hydrogen, Argon, and Carbon: can you vibe-code on Google alone?
Disclosure: No sponsorship or affiliate links in this article. Prices and plan access are from Google's public docs as of mid-Nov 2026. Benchmarks are lab- or leaderboard-reported as labeled. Oxygen and Helix are unverified rumors.
In September, Gemini 4 Argon landed as a frontier model most of us still couldn't use — Fairwind only. That was the wrong read on Google's pace.
Flash had already been sprinting. Then Argon went live for developers on 13 Oct, Carbon on 10 Nov, and Hydrogen on 12 Nov. The interesting question for vibe coding isn't "is Gemini competitive?" anymore. It's: if you only had Google — Antigravity, the Gemini app, and the Gemini API — would that be a solid setup?
TL;DR
- Google-only that works: Google AI Pro (~$20) + Antigravity. Hydrogen for everyday builds, Argon for hard planning, Carbon on Ultra (or when Pro quotas die).
- Free Antigravity is still on 3.8 Flash. Hydrogen is free in AI Studio with rate limits.
- "Google-only" means Gemini on purpose. Antigravity Pro also lists Claude Sonnet/Opus 5.5 — that's the escape hatch.
- API: Hydrogen $1.50 / $7.50. Argon and Carbon intro $2 / $10 through 30 Nov, then $4 / $20.
The Gemini 4 roster
| Model | Job | Price (per 1M tokens) |
|---|---|---|
| Hydrogen | Build, test, iterate | $1.50 / $7.50 (free tier in AI Studio) |
| Argon | Plan + most hard work | $2 / $10 intro → $4 / $20 after 30 Nov |
| Carbon | When Argon is stuck | Same intro as Argon · Ultra in Antigravity |
| Antigravity | Google's agent IDE | Pick Hydrogen / Argon / Carbon by plan |
Official posts: Argon, Carbon, Hydrogen, Antigravity models.
Google claims Hydrogen edges Argon on the Artificial Analysis Intelligence Index at high thinking (51 vs 50; 54 at xhigh). That does not mean Hydrogen replaces Argon for every architecture problem — Google still puts Argon and Carbon on deep judgment.
Prices and a few coding signals
| Signal | Hydrogen | Argon | Carbon |
|---|---|---|---|
| AA Intelligence Index | 51 high · 54 xhigh | 50 | ~59 at Carbon announce |
| CursorBench 4.0 (best) | — | — | 60.4% · $9.22 High |
| Terminal-Bench 4.1 | 57% vs 18.2% on 3.8 Flash | — | — |
CursorBench figure is from Cursor's public board (also cited next to Fable 5.2). FrontierCode for Carbon on Anthropic's Fable table was 41.1% at Medium — Anthropic-reported, not Google's.
Takeaway: Carbon is a CursorBench bargain at intro prices. Argon is Google's "majority of software engineering" frontier model. Hydrogen is the cheap, fast execution layer.
Which plan gets which model
Antigravity
| Model | Free & AI Plus | AI Pro (~$20) | AI Ultra ($100 / $200) |
|---|---|---|---|
| Gemini 3.8 Flash | Yes | Yes | Yes |
| Gemini 4 Hydrogen | No | Yes (paid non-trial) | Yes |
| Gemini 4 Argon | No | Yes (paid / eligible device trial) | Yes |
| Gemini 4 Carbon | No | No | Yes |
| Claude Sonnet / Opus 5.5 | No | Yes (paid non-trial) | Yes |
Gemini app / Search (rough)
- Free: Hydrogen via auto selection when it's the right pick.
- AI Plus (~$5): Hydrogen + higher app limits — not an Antigravity upgrade.
- AI Pro: Argon + Hydrogen.
- AI Ultra: Carbon + Argon + Hydrogen at the highest limits.
API / AI Studio
Hydrogen: free + paid. Argon and Carbon: paid only.
A Google-only workflow
Constraint: no Cursor, no Claude Code, no Codex as the daily path — only Google surfaces.
$0
- AI Studio / API: Hydrogen with rate limits.
- Free Antigravity: still 3.8 Flash. Learn the product; it's not Gemini 4.
~$20 Google AI Pro
- Hydrogen in Antigravity for everyday features, fixes, tests, deploys.
- Argon for architecture, auth, payments, multi-file hard judgment.
- Same model ids in AI Studio / API when you want a playground or a script.
- Stay off Claude in the picker if you're testing the Google-only thesis.
~$100+ Ultra
- You need Carbon, or Pro's five-hour / weekly limits keep dying.
This is editorial framing from Google's docs — not a week of production-only Antigravity logs. The stack is coherent on paper: Hydrogen builds, Argon plans, Carbon escalates.
Cross-vendor ladder (VCL framing)
Google doesn't publish this table. I do — as a roles map for vibe-coding jobs, not a claim that Helix equals Luna. Helix stays rumored.
| Role | Google (Gemini 4) | OpenAI | Anthropic |
|---|---|---|---|
| Cheap / fast / lite | Helix (rumor) | Luna | Haiku |
| Daily build | Hydrogen | Terra | Sonnet |
| Plan + hard judgment | Argon | Sol | Opus |
| When stuck | Carbon | Astra | Fable |
Rumors: Oxygen and Helix
Neither is on Google's public Gemini 4 list.
- Oxygen — Arena routing chatter mid-Nov; possible next Flash-class name. Unverified.
- Helix — reported Flash-Lite mapping in a deleted personal project. Unverified.
Until something ships, free Antigravity stays on 3.8 Flash, and AI Studio's free Hydrogen is the cheap Gemini 4 on-ramp.
FAQ
Can you vibe-code on Google alone?
Yes on Pro + Antigravity with Gemini models. Free Antigravity is still Flash. Ultra unlocks Carbon.
Is Hydrogen free?
In AI Studio / API, yes (rate-limited). In Antigravity, you need paid Pro (non-trial) or Ultra.
Do I need Ultra?
Only for Carbon or when Pro quotas aren't enough.
When do Argon/Carbon leave intro pricing?
After 30 Nov 2026: $4 / $20.
Full plan matrix, comparison notes, and related reading: vibecoderslife.com/post/gemini-4-google-only-vibe-coding.