Anzoth
Log inGet started
AnzothCycles · Gates · Flow

Models

Two public model IDs,
routed through Anzoth.

Call OpenAI-compatible model IDs through a private gateway with scoped API key permissions, usage visibility, RAG, and governed tool access.

Model availability varies by account and API key permissions.

Public models

Two stable model IDs are available today.

Anzoth-Core
Anzoth-Core
Available

Best for

General-purpose inference for reasoning, knowledge workflows, RAG-enabled applications, and customer-facing agents.

Reasoning, writing, and everyday chat workloads
RAG-enabled apps and knowledge assistants
Customer workflows that need a stable public model ID

Customer fit

OpenAI-compatible API
Scoped API key permissions
RAG and governed tools
Usage visibility per request
General purposeStable ID
Anzoth-Coder
Anzoth-Coder
Available

Best for

Coding workflows for software development, debugging, code assistance, and day-to-day technical work.

Software implementation and refactors
Debugging, review, and code explanation
Developer-focused tasks that benefit from precise responses

Customer fit

OpenAI-compatible API
Scoped API key permissions
RAG and governed tools
Usage visibility per request
Developer focusedStable ID

The public catalog currently includes Anzoth-Core and Anzoth-Coder. Model IDs stay stable even as the serving setup evolves.

Capabilities

What the gateway supports.

Capabilities are enabled at the gateway level and controlled per API key. Not all capabilities require model changes.

CoreChat completions
Available

Standard `/v1/chat/completions` endpoint. OpenAI-compatible request and response shapes. Streaming is supported via `stream: true`.

RAGRAG-grounded answers
Available

Connect a knowledge base to your account or API key. The gateway injects retrieved context before the model call and returns citations in metadata.

ToolsTool-governed workflows
Available

Governed tool access — web search, GitHub, file read/write — can be enabled per API key. Tools run through the gateway, not directly on the model.

ContextLong-context work
Available

Extended context windows are available on supported configurations. Contact support if you need long-context access for your account.

DevCoding assistance
Available

Code generation, explanation, and review via the standard completions interface. RAG can be layered with your own codebase documentation.

SupportCustomer support agents
Available

Build support bots that answer from your product docs. Per-customer API key permissions mean each tenant only sees their authorized knowledge base.

AccessMulti-model routing
Available

Route different API keys to different public models or tool permissions while keeping the model IDs stable in your app.

Routing

What happens when you call the API.

Every request passes through the Anzoth gateway before reaching the model. Authentication, model selection, RAG, tools, and usage visibility all run at the gateway layer.

Step 1
API key permissions

The gateway validates the bearer token, resolves the account, and checks which models and tools the key can use.

Step 2
Model selection

Your request selects a public model ID such as Anzoth-Core or Anzoth-Coder. Access is enforced before the request proceeds.

Step 3
RAG context

If a knowledge base is connected, retrieved context is added before the model call so answers stay grounded in your content.

Step 4
Governed tools

If governed tools are enabled, the gateway handles tool calls — web, file, code — and loops responses back through the model.

Step 5
Usage visibility

Input tokens, output tokens, model, latency, RAG usage, and tool calls are recorded per key and per account.

Example request

One model ID.
Everything routed.

Use Anzoth-Core in any OpenAI-compatible client. The gateway handles API key permissions, RAG, tools, and usage visibility before the model sees the request.

No SDK changes required
RAG layers in automatically when connected
Usage tracked per request, per key
Model ID is stable as backend evolves
example request · Anzoth-Core
curl
curl https://api.anzoth.com/v1/chat/completions \
-H "Authorization: Bearer anz-••••••••••••" \
-H "Content-Type: application/json" \
-d '{
"model": "Anzoth-Core",
"messages": [{
"role": "user",
"content": "Explain the routing layer."
}]
}'
POST /v1/chat/completions200 OK

Access

Public model access and permissions

The public catalog currently includes Anzoth-Core and Anzoth-Coder. Access is governed by your account and API key permissions, while the model IDs stay stable as the serving setup evolves.

View API examples

Start calling the API.

Create an account, generate a key, and your first request is minutes away.