API Reference

OMM exposes three compatible API protocols on a single server. Drop in any existing tool or SDK that uses Ollama, OpenAI, or Anthropic APIs without code changes.

Starting the Server

Terminal
bash
# Start on default port 11434
omm serve


# Custom host and port
omm serve --host 0.0.0.0 --port 8080


# With authentication
omm serve --api-key sk-your-secret-key

Ollama-Compatible API

Full compatibility with the Ollama REST API. Use any Ollama client library or tool with OMM as the backend.

Generate

POST/api/generate
Request
json
{
  "model": "llama3.2",
  "prompt": "Explain quantum computing in one paragraph",
  "stream": true,
  "options": {
    "temperature": 0.7,
    "top_p": 0.9,
    "num_predict": 256
  }
}

Chat

POST/api/chat
Request
json
{
  "model": "llama3.2",
  "messages": [
    {"role": "system", "content": "You are a helpful assistant."},
    {"role": "user", "content": "Hello!"}
  ],
  "stream": true
}

List Models

GET/api/tags
Response
json
{
  "models": [
    {
      "name": "llama3.2:latest",
      "size": 2019393189,
      "modified_at": "2026-04-28T10:00:00Z"
    }
  ]
}

Pull Model

POST/api/pull
Request
json
{
  "name": "qwen2.5:7b",
  "stream": true
}

OpenAI-Compatible API

Full compatibility with the OpenAI Chat Completions API. Use the OpenAI Python or Node SDK with a base URL change.

Chat Completions

POST/v1/chat/completions
Python SDK
python
from openai import OpenAI


client = OpenAI(
    base_url="http://localhost:11434/v1",
    api_key="omm"  # any non-empty string works without auth
)


response = client.chat.completions.create(
    model="llama3.2",
    messages=[
        {"role": "user", "content": "Write a haiku about coding"}
    ],
    stream=True
)


for chunk in response:
    if chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="")

Models Endpoint

GET/v1/models
Response
json
{
  "data": [
    {
      "id": "llama3.2",
      "object": "model",
      "owned_by": "local"
    }
  ]
}

Anthropic-Compatible API

Compatibility with the Anthropic Messages API. Use the Anthropic Python or Node SDK with a base URL change.

Messages

POST/anthropic/v1/messages
Python SDK
python
import anthropic


client = anthropic.Anthropic(
    base_url="http://localhost:11434/anthropic",
    api_key="omm"
)


message = client.messages.create(
    model="llama3.2",
    max_tokens=1024,
    messages=[
        {"role": "user", "content": "Explain recursion in simple terms"}
    ]
)


print(message.content[0].text)

Authentication

When started with --api-key, OMM requires an Authorization header on all requests:

Terminal
bash
# Ollama API
curl -H "Authorization: Bearer sk-your-secret-key" \
  http://localhost:11434/api/tags


# OpenAI API
curl -H "Authorization: Bearer sk-your-secret-key" \
  http://localhost:11434/v1/models


# Anthropic API
curl -H "x-api-key: sk-your-secret-key" \
  http://localhost:11434/anthropic/v1/messages

Streaming

All three APIs support Server-Sent Events (SSE) streaming when"stream": true is set in the request. Each chunk contains a partial response. The stream ends with a[DONE] message (OpenAI format) or a final JSON object with "done": true(Ollama format).

Model Parameters

Pass generation parameters per-request via the options or parameters object:

Temperaturetemperaturetemperature0.7Top Ptop_ptop_p0.9Top Ktop_k—40Max tokensnum_predictmax_tokens4096Repeat penaltyrepeat_penaltyfrequency_penalty1.1Stop sequencesstopstop[]Seedseedseed(random)
PreviousGPU SetupNextConfiguration