Gemini 4 Argon is Google’s newest frontier model, available worldwide since 13 October 2026. Google announced a second Gemini 4 model, Carbon, on 28 October, but only a small group of cyber defenders can use it so far. This guide explains the current Gemini lineup, what each model costs and which one suits which job.
What Gemini 4 Argon is and who can use it
Argon was announced on 30 September for trusted cyber defenders in Google’s Fairwind Program. Its global rollout began on 13 October. It is built for long, complex work such as software engineering, legal and finance tasks, and cybersecurity.
Argon is available in these places:
- Developers: the Gemini API paid tier, in Google AI Studio and Android Studio.
- Businesses: Gemini for business, the API and Google Antigravity, Google’s coding tool.
- Individuals: the Gemini app and AI Mode in Search, for Google AI Pro and Ultra subscribers.
Through the API, Argon handles up to 1 million tokens of input and output combined. Google reports that Argon ranks first on the Vals Index, a benchmark of finance, coding, legal and tax work.
What Gemini 4 Argon costs
Argon launched at an introductory API price. The price doubles after 30 November 2026.
| Period | Input (per million tokens) | Output (per million tokens) |
|---|---|---|
| Until 30 November 2026 | US$2 | US$10 |
| After 30 November 2026 | US$4 | US$20 |
Cached input tokens cost 95% less than the input price. If you plan to build on Argon, budget for the higher price, not the launch price.
What Gemini 4 Carbon is
Gemini 4 Carbon is Google’s strongest model, built for cybersecurity and long-horizon tasks. These are tasks where a model works through many steps on its own, sometimes for days. Carbon ranked first on version 5.0 of the Artificial Analysis Intelligence Index, an independent ranking, with a score of 59.
Carbon is not generally available yet. Access goes to a subset of cyber defenders first, then more trusted partners through the Fairwind Program. Google plans to release Carbon to developers, businesses and consumers “as soon as possible”, at the same introductory price as Argon.
The Gemini 3.x releases in 2026
Before Gemini 4, Google updated Gemini 3 in frequent steps. It released three Flash models in six weeks between July and September. The main releases:
| Date | Model | What it added |
|---|---|---|
| 12 February | Gemini 3 Deep Think upgrade | A stronger reasoning mode in the Gemini app for Google AI Ultra |
| 19 February | Gemini 3.1 Pro (preview) | More than double Gemini 3 Pro’s score on the ARC-AGI-2 reasoning test |
| 26 February | Nano Banana 2 | Faster image generation, from 512px up to 4K |
| 3 March | Gemini 3.1 Flash-Lite | The fastest, lowest-cost Gemini 3 model, for high volumes |
| 19 May | Gemini 3.5 Flash | Google’s strongest agent and coding model at the time |
| 19 May | Gemini Omni | Video creation from text, image and video prompts |
| 21 July | Gemini 3.6 Flash and 3.5 Flash-Lite | A new everyday model and a faster low-cost model |
| 13 August | Gemini 3.7 Flash | Better coding, and the model behind Gemini Spark, Google’s personal agent |
| 2 September | Gemini 3.8 Flash | The latest Flash model |
Gemini 3.8 Flash costs US$0.75 per million input tokens and US$3.75 per million output tokens. That introductory price ends on 31 December 2026. From 1 January 2027, it rises to US$1.50 and US$7.50. Budget for the higher price if you plan to use it beyond the end of the year.
Why fast model updates matter for your tools
New Flash versions now arrive every few weeks. A newer model can give different answers to the same prompt. If a business process depends on consistent output, test it again whenever you switch model versions. Also check the model name your tools call, so you know which version you are paying for.
Which Gemini model to use
This is our general guidance. Test with your own tasks before you commit.
- Gemini 4 Argon for complex reasoning, long documents, analysis and multi-step agent work.
- Gemini 3.8 Flash for everyday tasks where speed and cost matter, such as summaries and first drafts.
- Gemini 3.5 Flash-Lite or 3.1 Flash-Lite for very high volumes of simple work, such as tagging or sorting.
- Nano Banana image models for creating and editing images.
What is still to come from Google
- Gemini 3.5 Pro: in May, Google planned to release 3.5 Pro the following month. By July, 3.5 Pro was testing with partners. There is no release date yet.
- Gemini 4 Carbon: wider access “as soon as possible”.
- Genie 4: a world model for real-time interactive simulation, announced on 27 October. It is with a limited group of trusted testers. Access for Google AI Ultra subscribers and early access partners in the Gemini API is coming soon.
Google’s own October recap covers Argon, Carbon and Genie 4 together. Gemini also changed how people work inside Gmail, Docs and Sheets this year. We cover that in How Gemini changed work in Google Workspace.
References
14 official sources
- Gemini 4 Argon: Built to solve your hardest problems
- Gemini 4 Argon: our next era of frontier intelligence
- Gemini 4 Carbon: our next frontier model
- The latest AI news we announced in October 2026
- Introducing Gemini 3.8 Flash and 3.8 Flash Cyber
- Introducing Gemini 3.7 Flash
- Introducing Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber
- Gemini 3.5: frontier intelligence with action
- The Gemini app becomes more agentic, delivering proactive, 24/7 help
- Gemini 3.1 Flash-Lite: Built for intelligence at scale
- Nano Banana 2: Combining Pro capabilities with lightning-fast speed
- Gemini 3.1 Pro: A smarter model for your most complex tasks
- Gemini 3 Deep Think: Advancing science, research and engineering
- We're announcing Genie 4, our new world model for real-time interactive simulation