Anzoth
Log inGet started

Documentation

Build with the
API.

Use an OpenAI-compatible API for private inference, customer-scoped keys, RAG knowledge bases, and governed tool access.

Static preview docs. Live account provisioning is not enabled yet.

Quickstart

From zero to first response.

STEP 01

Create an API key

Sign up for an account. From the dashboard, navigate to API Keys and create a new key. You will see the full key once — copy it to your environment.

anz-live-sk · anz-3xK...9mP
STEP 02

Set your base URL

Replace the OpenAI base URL with the endpoint in your client configuration. No other request shape changes are required.

https://api.anzoth.com/v1
STEP 03

Send your first request

Make a chat completion call with the model ID anzoth-default. The API returns an OpenAI-compatible response object you can process immediately.

200 OK · {"choices":[…]}
first request · curl
curl https://api.anzoth.com/v1/chat/completions \
-H "Authorization: Bearer anz-••••••••••••" \
-H "Content-Type: application/json" \
-d '{
"model": "anzoth-default",
"messages": [{
"role": "user",
"content": "Hello from "
}]
}'
POST /v1/chat/completions200 OK
python · openai sdk
from openai import OpenAI
client = OpenAI(
api_key="anz-••••••••••••",
base_url="https://api.anzoth.com/v1"
)
response = client.chat.completions.create(
model="anzoth-default",
messages=[{"role": "user", "content": "Hello from "}]
)
print(response.choices[0].message.content)
openai>=1.0 · drop-in compatiblepip install openai

Compatible clients

Works with your existing SDK.

is OpenAI-compatible. Change one line in your configuration and your existing client code works without modification.

Python
openai>=1.0
JavaScript
openai npm package
LangChain
ChatOpenAI class
curl
any HTTP client
LlamaIndex
OpenAILike provider
REST
direct HTTP/1.1

Topics

Everything you need to build.

Getting started

Quickstart

Create your API key, point your client at the endpoint, and send your first chat completion in under five minutes.

Reference

API reference

Full reference for chat completions, models, usage, and errors. OpenAI-compatible request and response shapes.

Authentication

API keys

Every request is authenticated with a customer-scoped API key. Keys can be scoped, rate-limited, and revoked from your dashboard.

RAG

Knowledge bases

Upload documents, ingest them into a knowledge base, and connect them to your model. Hybrid dense+sparse retrieval, per-customer isolation.

Routing

Request routing

Requests pass through the gateway — authentication, model routing, RAG injection, tool orchestration, and usage recording happen before the model call.

Security

Governance

All traffic runs through -controlled infrastructure. Keys are hashed at rest, usage is logged per-request, and tenant RAG data is strictly isolated.

Error reference

401
UnauthorizedInvalid or missing API key. Check the Authorization header format: Bearer anz-…
403
ForbiddenThe API key does not have access to this model or endpoint. Check key scopes in your dashboard.
429
Rate limit exceededYour key has hit its RPM or token-per-minute limit. Back off and retry with exponential delay.
500
Internal errorAn unexpected error occurred in the gateway. This is an -side issue. Check status.anzoth.com.
502
Model unavailableThe model backend did not respond. The model may be warming up or temporarily unavailable.
503
Capacity unavailableAll inference capacity is currently in use. Retry with backoff. Contact support for persistent issues.

Start building.

Create an account to get your API key and start calling the API.