MoxieMoxie // Docs
M.1 STABLEOpen Terminal

Overview

Moxie provides a high-performance REST API for interacting with our local multimodal models. Our infrastructure supports standard chat completions with native vision capabilities, offered entirely for free via the DevSpace cluster.

Base URL
/api/chat
Authentication
x-api-key header

Model Selection

Moxie allows you to specify the inference engine based on your specific task requirements.

IDArchitectureBest For
moxie-qwen-vlQwen 2.5 (Multimodal)Vision, UI Debugging, Complex Logic
moxie-deepseek-v3DeepSeek V3 (MoE)Fast Coding, Math, Documentation Research

Chat Completions

Send a POST request to the completions endpoint to generate a response.

POST/api/chat

Request Body

{
  "model": "moxie-qwen-vl",
  "messages": [
    {
      "role": "user",
      "content": "Analyze this Expo dependency error."
    }
  ],
  "stream": true,
  "image": "optional_base64_string"
}

CURL EXAMPLE

SHELL
curl -X POST /api/chat/api/chat \
  -H "Content-Type: application/json" \
  -H "x-api-key: moxie-your-key-here" \
  -d '{
    "model": "moxie-deepseek-v3",
    "messages": [
      { "role": "user", "content": "Fix my package.json" }
    ],
    "stream": false
  }'

Vision Processing

To process visual data, encode your image as a Base64 string and include it in the message content using the standard multimodal format.

Pro Tip

Use moxie-qwen-vl for higher precision in OCR (reading error logs) and object detection in technical screenshots.