Octomil API

Welcome to the Octomil API. Use the switcher below to choose the right surface for your use case.

Local Inference (OpenAI-compatible)

• Base URL: http://localhost:8080/v1
• Use for: prompt/response inference on local or edge runtimes
• Primary endpoint: Inference

Start locally:

octomil serve phi-4-mini

Control Plane (Models/Devices/Rollouts)

• Base URL: https://api.octomil.com/api/v1
• Use for: model lifecycle, deployments, device management, rollouts, federated ops
• Start here: API Reference

Authentication:

curl https://api.octomil.com/api/v1/status \
  -H "Authorization: Bearer sk_live_your_api_key_here"

Get Started

If you're just getting started, follow the quickstart.

Libraries

Octomil local inference is compatible with OpenAI client libraries.

Python

pip install openai

import openai

client = openai.OpenAI(
    base_url="http://localhost:8080/v1",
    api_key="not-needed",
)

JavaScript

npm install openai

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "http://localhost:8080/v1",
  apiKey: "not-needed",
});

Control Plane Details

Authentication

All control-plane endpoints (except device registration) require Authorization: Bearer <token>.

Store your API key securely

Device API keys are returned once during registration and cannot be retrieved later.

Rate Limiting

Authenticated control-plane endpoints are rate-limited to 100 requests per minute per device.

Error Shape

{
  "error": "bad_request",
  "message": "Validation failed: 'device_id' is required.",
  "status_code": 400
}

Code	Error	Description
400	bad_request	Invalid or missing request fields
401	unauthorized	Missing or invalid API key
404	not_found	Resource does not exist
409	conflict	Resource already exists or state conflict
429	rate_limited	Too many requests; check Retry-After
500	internal_error	Unexpected server error

OpenAPI

Download openapi.yaml