Product
AssistantKnowledge BaseAcademyAI ModelsConnectorsAdministrationTokens & QuotasTrust
HostingGDPRPrivacySustainabilityIndustries
All industriesLaw FirmsAccountants & CPAsManufacturingHealthcareReal EstateConstructionInsurance & FinanceRetailHospitality & TourismPublic SectorPricing
Pricing overviewCalculatorResources
VideosDownloadFAQBook a demoLanguage
The right engine for every task, on our servers.
emi runs open-source AI models on our dedicated infrastructure. Pick the mode that fits your task — or let emi decide automatically.
CHOOSE MODE
The basics, simply put
A model is the engine behind the conversation
emi is the workspace: where you write, store documents and manage permissions. The model is the “brain” that understands your prompt and writes the reply. Just like an engine, it can be upgraded without changing the car.
emiYour workspace, your documents, your access permissions.Why multiple models?
Not every task is equal
Rewriting a single sentence doesn’t take the same effort as reviewing a forty-page contract.
Smarter quota usage
Smaller models consume fewer tokens — saving the heavyweights for your toughest problems.
Lower energy footprint
Matching the model to the task also means avoiding wasted compute.
emi’s four modes
Speed, depth, consumption: dialed to your needs
A qualitative comparison so you can tell them apart at a glance.
Automatic
emi picks the right mode based on your request. The best default choice.
Fast
Short emails, rewrites, quick translations and direct questions.
Standard
Summaries, meeting notes, outlines and most everyday tasks.
Deep Thinking
Contracts, multi-document comparisons and multi-step reasoning.
Illustrative gauges, not benchmark figures. Your administrator chooses which modes are available to your team.
Which mode for my task?
Pick a task and see which mode we recommend
RECOMMENDED MODE FOR “A short email”
A few lines and the right tone: a lightweight model answers in a blink and uses very little compute.
In Automatic mode, emi makes this choice for you.
In practice
One click before hitting send — that’s all
The mode selector sits right next to the message bar. Switch modes from one question to the next inside the very same conversation.
Why open source
Open models running on our infrastructure, not a Big Tech cloud.
Most AI assistants send your prompts out to a giant provider’s cloud, often in the US. emi does the opposite: the models come to us and run on our dedicated servers.
Transparency
Open-source models are published openly: we know exactly what runs under the hood.
Zero lock-in
No forced dependency on a US tech giant, its price hikes or shifting terms of service.
Dedicated execution
Models run directly on our dedicated servers.
Your data stays here
Nothing is sent to a third-party model vendor, and nothing ever trains an AI.
Always up to date
When a better open model comes out, we can adopt it without your team changing tools.
One consistent interface
Your teams only ever see emi. Engines evolve while daily habits stay intact.
Sample model families
Examples of open-source model families that can run on our infrastructure. The exact lineup offered inside emi evolves over time.
Fully open Swiss sovereign model launched in 2025.
Such as Mistral Small or Ministral.
A broad family ranging from compact to large models.
Open-weights models released publicly.
Open models engineered for deep reasoning.
Questions about AI models
Do I have to pick a mode before every question?
No. Automatic mode picks the right fit for you. Manual modes are simply there whenever you want direct control.
Are open-source models less capable?
Leading open models excel at everyday enterprise work: drafting, summarizing, translating and analyzing documents — and we can adopt newer releases as soon as they raise the bar.
Who decides which models are enabled?
Your workspace administrator chooses which modes are available to your team inside AI Configuration.
Capable AI without your data ever leaving
Compare modes live on your own tasks during a 30-minute demo.