Dual-Model API

🐾 Recommended for OpenClaw

Two Top Models,
One API Call

Opus handles reasoning, planning, and code review. GPT handles coding and execution. Together, they outperform any single model. Just swap the endpoint.

Opus 4.6

Reasoning

orchestrate

GPT-5.4

Coding

Best Use Case

The Strongest Backend for OpenClaw

OpenClaw needs an LLM backend to power agent workflows. With AI Duet, Opus makes decisions and reviews, GPT codes and executes. Two models in concert produce better output than either alone.

🔌

Swap the Endpoint

Replace the API URL in your OpenClaw config with api.aiduet.app and enter your AI Duet key.

🤖

Dual-Model Duet

Every instruction is analyzed by Opus, coding tasks are delegated to GPT at full speed. Each model does what it does best.

✅

Auto Review Loop

Opus reviews every line of GPT's output. Doesn't pass? Goes back to GPT. Code review by the strongest reasoning model, built in.

openclaw.config

api_url = "https://api.aiduet.app"
api_key = "hr-your-key-here"
model   = "opus-4.6"

Core Advantages

Why AI Duet

🎯

Best Reasoning + Fastest Coding

Opus is the strongest reasoning model. GPT is the fastest coding model. AI Duet makes them collaborate — plan, code, review in one flow. No single model can do this.

🔄

Built-in Code Review

Opus reviews every line of GPT's output. Doesn't pass? Sent back for revision. Like having an Opus-level code reviewer on every request.

⚡

Dual Format

Anthropic Messages API and OpenAI Chat API both work. OpenClaw, Cursor, your own system — just swap the endpoint.

🛡️

Real-time Streaming

SSE streaming delivers Opus thinking and GPT code output in real time. Same experience as a direct API connection.

📊

Cheaper Than Official

Every token is 20% below official API rates. Pay per token, no subscriptions. Real-time usage breakdown in your dashboard.

🔌

One-line Setup

Swap the API endpoint and key. Works with OpenClaw, Cursor, Continue, Cline, and all major AI coding tools.

How It Works

One Call, Dual-Model Duet

Opus Plans

Opus receives the request, understands intent, and breaks it into executable coding instructions. The strongest reasoning model does what it does best.

GPT Codes

GPT takes over and generates code at full speed. The fastest coding model does what it does best.

Opus Reviews

Opus reviews every line of code. Passes? Final response. Doesn't pass? Back to GPT. Built-in quality gate.

Pricing

Pay As You Go

No monthly fees. No hidden costs. Pre-load balance, pay per token.

Find and fix a login validation bug

~50K input / 20K output

Pure Opus

$0.75

Cost per request

Opus handles everything alone

Save 41%

AI Duet

$0.44