Every major AI model. One subscription. Hosted in Switzerland.
GPT, Claude, Gemini, Mistral and the rest in a single interface — instead of stacking five subscriptions. Your account and conversations stay in Switzerland.
The service opens soon. Sign up to hear first.
The models
Switch models, not subscriptions
One model stuck? Move to the next one inside the same conversation. Every model shows where your request is processed.
The waitlist
What you get by signing up
Early access
You are among the first to open an account, before the public launch.
Launch price locked in
The price shown when you sign up stays yours, even if it rises later.
No spam
One announcement, at launch. One-click unsubscribe, any time.
Your input counts
The plan you pick directly shapes what we build first.
Features
What you will be able to do
One interface to write, analyse, search and create — with the model that suits each task.
Every model
Switch from one model to another inside a conversation, without starting over.
Projects
Group conversations, documents and instructions by folder, so your context is where you left it.
Documents and files
Drop in a PDF, a spreadsheet or a contract and question it directly.
Image generation
Create and edit visuals without leaving your conversation.
Web search
Answers grounded in recent sources, with the links to check them.
History on your terms
Search your past conversations — or delete them for good.
Data sovereignty
Your data stays in Switzerland. The models come from everywhere.
Your account, conversations and files are stored in Switzerland with Infomaniak. Your requests, though, are processed by the provider of whichever model you choose — and every model shows its processing region.
Swiss datacenters
Infrastructure owned and run by Infomaniak, with no intermediary.
ISO 27001 certified
Along with ISO 9001, 14001 and 50001: quality, environment, energy.
Swiss law
Beyond the reach of extraterritorial laws such as the US CLOUD Act.
No training on your data
The professional APIs we use exclude training on the data you send.
Region per model
Switzerland, European Union or United States: shown per model, and filterable.
Renewable energy
Datacenters powered by renewable energy, with heat reuse.
The maths
What you pay today, separately
Subscribing to each service individually adds up fast — and leaves you juggling five interfaces.
- SuperGrokCHF 30.–
- ChatGPT PlusCHF 23.–
- Google AI ProCHF 22.–
- Claude ProCHF 20.–
- Le Chat ProCHF 15.–
Providers’ public prices, noted for reference.
Privacy
What we do not do
No training on your data
Not by us, and not by the model providers: their API terms exclude it.
Deletable history
Erase a single conversation or your entire history whenever you want.
Encrypted in transit and at rest
TLS 1.3 for traffic, encryption for stored data.
No data broking
We live on subscriptions. Your data is never sold or shared for advertising.
Questions
What people ask us
Which models will be available?
At launch we plan to cover the four major families:
- GPT-5.1 (OpenAI) — writing, analysis, general use
- Claude Sonnet 5 and Claude Opus 5 (Anthropic) — long documents, code, reasoning
- Gemini 3 Pro (Google) — multimodal, extended context
- Mistral Large (Mistral AI) — the only one on the list processed in the European Union
Grok 4 (xAI) and DeepSeek V3 are under consideration and may be added depending on demand — say so when you join the waitlist.
You move between them inside a single conversation, without starting over. New versions are added as they ship, at no extra cost: that is precisely what you buy with a bundled subscription rather than five separate ones.
The final catalogue is published at launch, and every model shows the region where your request is processed.
Does my data really stay in Switzerland?
Your account, conversations, projects and files are stored in Switzerland. When you query a model, however, your request is passed to that model’s provider for processing — in the United States for OpenAI, Anthropic and Google, in the European Union for Mistral.
That is true of any service giving access to these models, including subscribing to them directly. We would rather write it down plainly and show the processing region for each model, so you can filter when your case requires it.
Are my conversations used to train the models?
No. We use the providers’ professional APIs exclusively, whose terms exclude training on the data sent to them. We do not train any model on your conversations either.
When does the service open?
We are not setting a public date yet, and we would rather say so than announce a deadline we would miss.
Concretely, what remains is finalising contracts with each model provider, completing the Swiss infrastructure rollout, and running a closed testing phase with a first group of users. That last step is what will set the date.
What we can promise: people on the waitlist are contacted before the public launch, and they are the ones who make up the closed testing group.
Is the pricing shown final?
No. The amounts shown are indicative and exist to give you a sense of the offer. Final prices are announced at launch.
If you are on the waitlist, the launch price stays yours, even if it rises afterwards.
How do I leave the waitlist?
Every email we send carries a one-click unsubscribe link. You can also write to
privacy@marmott.ai and ask us to delete your address — we delete it, no
questions asked.
I did not receive the confirmation email
It usually arrives in under a minute. If it does not:
- Check your spam folder — by far the most common cause.
- Check the spelling of your address: one wrong letter and the email goes somewhere else. If so, sign up again with the correct address.
- Some corporate mail systems hold messages for a few minutes.
Still nothing? Write to contact@marmott.ai and we will confirm your signup
manually.
How is this different from subscribing directly?
Three concrete differences.
Price. Subscribing separately to ChatGPT, Claude, Gemini and Mistral runs to roughly a hundred francs a month. A bundle pays for itself as soon as you use two models regularly.
Use. One interface, one history, one place to search. You switch models mid-conversation when one gets stuck, without re-explaining the context in another tab.
Hosting. Your account, conversations and files are stored in Switzerland — which none of the providers offer if you subscribe directly with them.
That said, here is what you do not get: each provider’s most specific features usually land in their own app first. If you use a single model intensively and want its newest capabilities on day one, subscribing directly remains the better fit.
Are there usage limits?
Yes, and we would rather say so plainly than write “unlimited” in bold with an asterisk.
Each plan carries a monthly usage allowance. Standard offers roughly three times the Starter allowance, and Expert roughly ten times. Exact volumes are published with the final pricing.
If you reach your limit your account is not blocked: you continue on a lighter model until renewal, or change plan mid-month.
In practice Starter covers steady daily use. The higher plans are aimed at image generation, large documents and reasoning models, which consume considerably more.
Ready to stop juggling five subscriptions?
Join the waitlist. We will tell you as soon as we open, and the launch price stays yours.