API Documentation & Integration Guide

High-performance LLM inference, OpenAI-compatible proxy, and Native Ollama APIs.

The gateway provides standard OpenAI-compatible (/v1) endpoints and Native Ollama (/api) endpoints with SHA-256 hashed Bearer token authentication, real-time chunked streaming, and hardware compute metering.

🔑 1. Authentication & API Key Pre-Fill

Select or generate an API key below to automatically populate your Bearer token in all code snippets on this page.

Authorization: Bearer pcw_live_secret_token_here

⚡ 2. Zed IDE Configuration

In Zed, press Cmd + , or open ~/.config/zed/settings.json and paste either provider configuration below:

Method A: OpenAI Compatible Provider (Requires Bearer Token)

{
  "language_models": {
    "openai": {
      "api_url": "https://api.proxyai.co.uk/v1",
      "available_models": [
        {
          "name": "ornith-1.5:35b",
          "display_name": "ornith-1.5:35b (Local GPU)",
          "max_tokens": 32768
        },
        {
          "name": "ornith-1.5:35b",
          "display_name": "Ornith 1.5 35B (Local GPU)",
          "max_tokens": 262144
        }
      ]
    }
  }
}

When Zed prompts for the OpenAI API Key in the UI, enter your token: pcw_live_secret_token_here

Method B: Native Ollama Provider

{
  "language_models": {
    "ollama": {
      "api_url": "https://api.proxyai.co.uk?key=pcw_live_secret_token_here",
      "available_models": [
        {
          "name": "ornith-1.5:35b",
          "display_name": "ornith-1.5:35b (Local GPU)",
          "max_tokens": 32768
        },
        {
          "name": "ornith-1.5:35b",
          "display_name": "Ornith 1.5 35B (Local GPU)",
          "max_tokens": 262144
        }
      ]
    }
  }
}

Base URL is https://api.proxyai.co.uk. Model discovery probes /api/tags and /api/version.

🚀 3. OpenAI SDK Quickstart (/v1)

Point standard OpenAI client libraries to https://api.proxyai.co.uk/v1:

cURL (Chat Completions)

curl -X POST "https://api.proxyai.co.uk/v1/chat/completions" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer pcw_live_secret_token_here" \
  -d '{
    "model": "ornith-1.5:35b",
    "messages": [
      {"role": "user", "content": "Explain quantum computing in 2 sentences."}
    ],
    "stream": false
  }'

Python (openai >= 1.0)

from openai import OpenAI

client = OpenAI(
    base_url="https://api.proxyai.co.uk/v1",
    api_key="pcw_live_secret_token_here"
)

response = client.chat.completions.create(
    model="ornith-1.5:35b",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Explain quantum computing in 2 sentences."}
    ],
    stream=True
)

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

Node.js / TypeScript (openai npm)

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.proxyai.co.uk/v1",
  apiKey: "pcw_live_secret_token_here",
});

const stream = await client.chat.completions.create({
  model: "ornith-1.5:35b",
  messages: [{ role: "user", content: "Hello AI Gateway!" }],
  stream: true,
});

for await (const chunk of stream) {
  process.stdout.write(chunk.choices[0]?.delta?.content || "");
}

🦙 4. Native Ollama API Endpoints

Native Ollama endpoints are served on https://api.proxyai.co.uk with NDJSON streaming:

curl -X POST "https://api.proxyai.co.uk/api/generate" \
  -H "Authorization: Bearer pcw_live_secret_token_here" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "ornith-1.5:35b",
    "prompt": "Why is the sky blue?",
    "stream": false
  }'

⚡ 5. Energy (kWh) Billing Formulation

For self-hosted GPU Ollama profiles, your account balance is deducted based on the actual physical compute power consumed by your query:

Energy (kWh) = (Power in Watts × Execution Duration in Seconds) / 3,600,000

Charge (£) = Energy (kWh) × Rate per kWh (e.g. £0.30/kWh)