One runtime.
Three ways to deploy.

Same ARK inference engine across the line — from zero-setup cloud API to fully air-gapped on-premises. Pick the control model that matches your risk posture, your operating preference, and the platform you already run.

Pick the ARK that fits you.

Three deployment modes, one codebase. Same models, same OpenAI v1-compatible API, same patch cycle — the only thing that changes is where it runs and who operates it.

Fastest start
ARK Cloud

Managed EU-sovereign API

Start in minutes. No GPUs. No infrastructure. Every token stays inside the EU.
Read more

Call frontier open-source models through an OpenAI v1-compatible endpoint hosted entirely in the EU. Zero retention by default, transparent per-token pricing, free credits to get started.

Hosted by
ARK Labs
Data residency
EU
Pricing
Per-token, self-service
SLA
99% best-effort
  • 10+ curated frontier models (Llama, Qwen, DeepSeek, GPT-OSS, Mistral)
  • Real-time API, Batch API, and a built-in Portal for non-technical teams
  • Zero retention, GDPR-ready, free credits on signup
  • Best for: developers, product teams, AI-native startups
Minimum footprint
ARK Core

Licensed runtime only

Just the inference engine. You run the rest of the platform. Maximum control, minimum vendor footprint.
Read more

The ARK runtime, Compute Nodes, Supervisor, API Gateway, and Model Storage — licensed for your stack and wired into the identity, logging, and monitoring you already operate. Same software as ARK Tailored; less scope.

Hosted by
You (cloud or on-prem)
Data residency
Your perimeter
Pricing
Annual runtime license
SLA
Software Support SLA
  • Inference engine + Compute Nodes, Supervisor, API Gateway, Model Storage
  • Attach your own Keycloak, ELK, Prometheus/Grafana, Splunk, Datadog
  • Air-gap supported; upgrade path to ARK Tailored is a config change
  • Best for: mature platform teams, sovereign & government, infra integrators
Full platform
ARK Tailored

Full platform on your infrastructure

Everything in ARK Core, plus the managed platform services enterprises expect — on your hardware, in your perimeter.
Read more

The full ARK Platform deployed inside your VPC or on-premises. Identity, logging, telemetry, monitoring, and Portal included and swappable with your existing stack. ARK deploys and configures; your team operates; we support the software.

Hosted by
You (cloud or on-prem)
Data residency
Your perimeter
Pricing
Annual subscription
SLA
Software Support SLA
  • Core runtime + managed platform (Identity, Logging, Telemetry, Monitoring, Portal)
  • Integrates with SAP, Salesforce, SharePoint, ServiceNow, Microsoft 365, Oracle
  • Air-gap supported; SSO, SAML, OIDC enterprise identity
  • Best for: regulated enterprises, AI SaaS platforms, sovereign cloud partners

Start here. Three quick questions.

Most customers land on one of these three patterns within a minute of reading them.

I want to start building today, no infrastructure.
You want ARK Cloud. Sign up, get free credits, call the OpenAI v1-compatible API from your existing SDK. Every token runs in the EU with zero retention.
I already run Keycloak, ELK, Prometheus — I just need the inference engine.
You want ARK Core. Licensed runtime only. You integrate it into your existing platform stack. If you need the managed services later, upgrading to ARK Tailored is a config change, not a migration.
Data must stay inside my perimeter, and I want the full platform managed for me.
You want ARK Tailored. Full ARK Platform deployed on your cloud or on-prem, with identity, logging, telemetry, monitoring, and Portal included. Configured to your standards, handed over to your team, Software Support SLA.

Side-by-side comparison.

Same runtime, same models, same API. Scope, hosting, and operating model are what differ.

  ARK Cloud ARK Core ARK Tailored
Hosting ARK-hosted (EU) Your infrastructure Your infrastructure
Operating model Fully managed You deploy and operate ARK deploys, you operate
OpenAI v1-compatible API
Core runtime (Compute, Supervisor, Gateway, Model Storage)
Identity, Logging, Telemetry, Monitoring Managed by ARK Bring your own Included (swappable)
Portal (chat & embeddings UI)
Air-gap / fully isolated deployment
Custom / fine-tuned models Roadmap
Pricing model Per-token, self-service Annual runtime license Annual subscription
Service level 99% best-effort Software Support SLA Software Support SLA
Best fit Developers, startups, product teams Mature platform teams, government, integrators Regulated enterprises, AI SaaS, sovereign cloud

Still deciding? Let's shape it together.

Tell us about your workload, your compliance regime, and the platform you already run. We'll map you to the right tier and size the deployment to the traffic you actually have.