Point existing tools at a governed endpoint.

Point existing software at OpenAI-compatible HTTP endpoints. The same routing, budgets, and evidence apply as for every other request.

Platform capabilities

  • OpenAI-compatible Chat Completions, Responses, text embeddings and multipart audio transcription, subject to provider support
  • Manual API keys for your active organisation. Members create inherited-budget keys; owners and admins also manage dedicated keys.
  • Explicit connection and model selection through the standard API
  • The same data-routing matrix, key floors, optional budgets and per-key request evidence

Who it's for

  • Automation and internal tools using supported OpenAI-compatible endpoints
  • Retrieval pipelines needing text embeddings with a pinned model and preprocessing identity
  • Speech workflows needing transcription and clip-relative word or segment timings
  • Applications that reuse organisation model choices while retaining their own user-access controls

API-specific visibility

  • Per-key attribution so you know which system used inference
  • The same D×E routing matrix as every other request, plus a key default and floor you cannot lower
  • Optional header to raise sensitivity per request without a custom request body

Full activity and administrative audit details are under Governance.

Ecosystem

Built for the AI ecosystem

Agents, automation, and custom AI systems connect to Steinkauz AI for the same models, routing, and evidence as every other request.

OpenCode

Open-source coding agent in the terminal.

n8n

Workflow automation for teams and internal tools.

Hermes

Agent runtime from Nous Research.

Paperclip

Orchestrate and run teams of AI agents.

And many more

Clients using the supported OpenAI-compatible endpoints can connect. Check their required features against the API documentation.

Build your own

Ship OpenAI-compatible systems on Steinkauz AI using the OpenAI SDK or any HTTP client as your foundation.

Quick start

Quick start

Create an API key in the platform, then call the endpoints with any OpenAI-compatible client.

  1. Use an active Steinkauz Cloud organisation or a Private Deployment.
  2. Add a provider connection, review Policies and Budgets, then create a key under API access.
  3. Set STEINKAUZ_MODEL to a complete language-model ID from your configured connection. Point your client at the Steinkauz AI base URL and send a chat completion request.
# Set STEINKAUZ_MODEL to a configured language-model ID.
# List available model IDs with GET /v1/models if the provider supports listing.
curl -sS "$STEINKAUZ_BASE_URL/v1/chat/completions" \
  -H "Authorization: Bearer $STEINKAUZ_API_KEY" \
  -H "Content-Type: application/json" \
  -d "{\"model\":\"$STEINKAUZ_MODEL\",\"messages\":[{\"role\":\"user\",\"content\":\"Hello\"}]}"

Build on a governed foundation

Use the same models, providers, and policies through OpenAI-compatible APIs.

Get StartedRead the docs