GME Specification

Overview

Global Models Endpoint (GME) is an open standard that defines a universal API format for AI providers. Any provider implementing GME can be integrated into applications using the same endpoints, request structure, and response format.

Base Endpoint

/gme/v1

Chat Endpoint

POST /gme/v1/chat

Request

{
    "model":"example-model",
    "input":"Hello!",
    "temperature":0.7,
    "max_tokens":512
}

Response

{
    "id":"chat_001",
    "model":"example-model",
    "output":"Hello!",
    "finish_reason":"stop",
    "usage":{
        "input_tokens":5,
        "output_tokens":2,
        "total_tokens":7
    }
}

Authentication

Providers should authenticate requests using the Authorization header.

Authorization: Bearer YOUR_API_KEY

Provider Compatibility

Providers implementing the required GME endpoints may identify themselves as GME Compatible.

Extensions

GME is an evolving standard. If you would like to propose a new endpoint, request field, response field, or any other extension, please send your proposal by email.

📧 yassinkam01@gmail.com

Approved extensions may be included in future versions of the GME Specification.