2kw.ai
Sign inBook a 15-min demo
Gateway

One OpenAI-compatible LLM gateway for every provider

Route OpenAI, Anthropic, Azure OpenAI, Google Vertex AI, and self-hosted Ollama models through a single endpoint. Switch providers by changing a config line — not your codebase.

Why route through a gateway?

OpenAI doubles their prices. Anthropic ships something better. A client demands on-premise models. If your integration is hard-wired to one provider, each of these events becomes a migration project. 2kw.ai sits between your application and the model providers: your code talks to one OpenAI-compatible API, and the gateway routes every request to the provider you configured.

Because the interface is OpenAI-compatible, existing OpenAI integrations work out of the box — point your SDK at a new base URL and keep going. Moving a workload from one model to another is a config change, not a refactor.

OpenAI-compatible endpoint
Use your existing OpenAI SDKs and tooling unchanged. One base-URL swap and requests flow through the gateway.
Every major provider
OpenAI, Azure OpenAI, Anthropic, and Google Vertex AI — plus self-hosted models via Ollama for on-premise requirements.
Switch by config
Move traffic between providers and models by changing a config line. Your integration code stays untouched.
Bring your own keys
On Scale and Enterprise, connect your existing provider credentials. Keys are encrypted at rest, and BYOK requests don't consume your plan's token quota.
Cost tracking built in
Every request is metered per token and per euro, with budgets and alerts that keep spending visible before the invoice arrives.
EU data residency
Built-in models run GDPR-compliant with EU data residency, operated by Manfred Kunze Development GmbH in Germany.

EU data residency and GDPR

For teams in regulated industries, where a prompt is processed matters as much as what it returns. 2kw.ai's built-in models run with EU data residency and GDPR compliance. If you bring your own keys instead, your data flows to the provider you chose under their terms — the trade-off stays explicit and per-workload rather than buried in your integration.

Works where you work

The gateway is reachable over a standard REST API, through an MCP server that plugs 2kw.ai directly into Claude Code, Cursor, and Windsurf, and via a CLI built for agentic workloads. Full reference and quickstarts live in the documentation.

More from the platform
Document extraction Schema-driven structured data extraction from 20+ document formats.Observability Every token, every euro, every field — budgets, alerts, and audit trails.

Stop hard-wiring your stack to one provider.

Start 7-day free trialBook a 15-min demo