Duck Proxy — API Command Center

Admin · API Key Management

Status & Port

18080 ● OK

http://127.0.0.1:18080/v1

Active Models

GPT-5, Claude, Mistral, Gemma

Protocol

OpenAI v1

Chat, Streaming, Images

Memory Footprint

< 13 MB

Zero-leak async runtime

API Command & Endpoint Tester

Usage Tokens (In / Out)

— / —

0 requests

Live Usage — 10 Request Terakhir

memuat…

GET /v1/models

Description

Queries the proxy for all available LLM models, routing aliases, and ownership flags.

Target URL

READY

// Click "Run Command" to fetch available models...

POST /v1/chat/completions

Model Identifier

Prompt Message

READY

// Click "Send Query" to execute completion...

POST /v1/chat/completions (SSE Stream)

Streaming Model

Streaming Prompt

READY

// Real-time token streaming will render here...

POST /v1/images/generations

Image Prompt

Uses Duck AI's image generator model and extracts base64 image data into standard OpenAI JSON format.

READY

// Generated image will appear here...

Available Model Catalog

All models are routed dynamically with isolated upstream sessions and automated V8 anti-bot solving:

Model	Provider	Primary Strengths & Best Use Cases
gpt-5.6-luna	OpenAI	Flagship Coding: Complex multi-file architecture, deep logic, autonomous reasoning.
claude-haiku-4-5	Anthropic	Fast Iteration: Code review, unit test generation, docstrings, interactive refactoring.
mistral-small-2603	Mistral AI	Speed & Algorithms: Mathematical logic, concise scripts, algorithms.
gemma4-31b	Google / Tin	Privacy Preserved: High-parameter open model with zero-tracking guarantees.
gpt-5.4-mini	OpenAI	Lightweight: Quick syntax lookups, commit message drafting, simple queries.
image-generation	Duck.ai	Visual Assets: Generates images, app icons, vector art via standard OpenAI format.

Universal IDE & Editor Integration Hub

Connect Duck Proxy to your favorite editor or tool in under 30 seconds:

1. Open Cursor Settings (Ctrl+Shift+J / Cmd+Shift+J) → Models.
2. Under OpenAI API Key, type duck-proxy.
3. Click Override OpenAI Base URL and enter http://localhost:18080/v1.
4. Add models: gpt-5.6-luna, claude-haiku-4-5, mistral-small-2603.

Base URL: http://localhost:18080/v1
API Key:  duck-proxy
Models:   gpt-5.6-luna, claude-haiku-4-5, mistral-small-2603

Add these models to your ~/.continue/config.json:

{
  "models": [
    {
      "title": "Duck GPT-5 Luna",
      "provider": "openai",
      "model": "gpt-5.6-luna",
      "apiBase": "http://localhost:18080/v1",
      "apiKey": "duck-proxy"
    },
    {
      "title": "Duck Claude Haiku",
      "provider": "openai",
      "model": "claude-haiku-4-5",
      "apiBase": "http://localhost:18080/v1",
      "apiKey": "duck-proxy"
    }
  ]
}

1. In the extension settings, choose OpenAI Compatible provider.
2. Base URL: http://localhost:18080/v1
3. API Key: duck-proxy
4. Model ID: gpt-5.6-luna or claude-haiku-4-5

Provider: OpenAI Compatible
Base URL: http://localhost:18080/v1
API Key:  duck-proxy
Model ID: gpt-5.6-luna

Run terminal pair programming with Aider:

export OPENAI_API_BASE="http://localhost:18080/v1"
export OPENAI_API_KEY="duck-proxy"
aider --model openai/gpt-5.6-luna

Add to your ~/.config/zed/settings.json:

{
  "language_models": {
    "openai": {
      "api_url": "http://localhost:18080/v1",
      "available_models": [
        { "name": "gpt-5.6-luna", "display_name": "Duck GPT-5 Luna", "max_tokens": 8192 },
        { "name": "claude-haiku-4-5", "display_name": "Duck Claude Haiku", "max_tokens": 8192 }
      ]
    }
  }
}

Settings → AI Provider → Custom OpenAI:

Endpoint: http://localhost:18080/v1
API Key:  duck-proxy
Model:    gpt-5.6-luna

Avante.nvim setup:

require('avante').setup({
  provider = "openai",
  openai = {
    endpoint = "http://localhost:18080/v1",
    model = "gpt-5.6-luna",
    api_key_name = "DUCK_PROXY_KEY",
    timeout = 30000,
  }
})

Standard OpenAI Python SDK integration:

from openai import OpenAI

client = OpenAI(
    base_url="http://localhost:18080/v1",
    api_key="duck-proxy"
)

response = client.chat.completions.create(
    model="gpt-5.6-luna",
    messages=[{"role": "user", "content": "Hello Duck Proxy!"}],
    stream=True
)

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

Direct terminal SSE streaming:

curl -N http://localhost:18080/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{"model": "gpt-5.6-luna", "stream": true, "messages": [{"role": "user", "content": "Hello!"}]}'

DUCKPROXY