OpenAI-compatible · Private · Governed
PRIVATE INFERENCE.
Your app connects. We handle routing, keys, RAG, and tools.
Drop-in compatible
Works with your existing code.
Anzoth's API is OpenAI-compatible. Point your SDK at our endpoint, swap your key, and go.
Compatible with any client using the
OpenAI Python SDK, openai-node, or direct HTTP.
Platform
Everything between your app and the model.
Request routing
Every call passes through our gateway. Authentication, rate limiting, model routing, and usage tracking — handled before your model sees a token.
Customer API keys
Issue per-customer keys with scoped permissions, token quotas, and RPM limits. Revoke or rotate at any time.
Governed access
Control which models each customer can call. Enable or restrict RAG, tools, and file access per key.
Knowledge bases
Your data.
In every answer.
Upload documents, connect a knowledge base, and your model answers with your content — not guesses.
- Hybrid dense + sparse retrieval
- Per-customer knowledge isolation
- Ingest status and citation tracking
3 documents · 14,802 chunks · Last indexed 2h ago
API Keys
One key. Full control.
Scoped permissions, spending limits, per-key rate controls, and last-used tracking.
Production
anz-live-sk · anz-3xK...9mP · Last used 4m ago
Built for private inference.
Client apps connect to our API gateway. The gateway handles authentication, routing, usage tracking, model access, RAG, and tool orchestration. Your traffic is routed through Anzoth-controlled infrastructure with customer-level access controls, usage tracking, and governed tool access.
Keys hashed, never stored plain
Tenant-isolated usage and RAG
Per-request audit logs
Rate limits enforced at gateway