Ship apps. Make videos.
Build a real business.

qcode is the AI coding agent that writes the code, deploys the app, renders the video, and bills your customers — all behind one key. Built for makers shipping fast.

latest · qcode desktop

Native app for your machine.

Full filesystem access. Spawn local builds, render Remotion videos to your machine, edit large repos. Your code never leaves your machine until you ask it to.

curl -fsSL https://qlaud.ai/install/qcode | bash

irm https://qlaud.ai/install/qcode.ps1 | iex

All platforms + checksums·Source on GitHub

Free · in your browser

qcode · chat

Draft a cold email to 20 AI founders…

claude-sonnet-4-6gpt-5pick any model

Chat with qcode in your browser.
Free. No install.

Open qcode Sign up free

API keys Connectors Model catalog Pricing

qcode · Claude Sonnet 4.6

Routable through one API · 14+ providers

ClaudeGPTGeminiGrokLlamaDeepSeekQwenKimiMistralGroqCerebras

Same API shape, hard spend caps enforced by the gateway, automatic provider failover.

Why qcode

Three things Cursor, Bolt, and Claude Code don’t do.

Hard spend caps

Your AI bill won’t surprise you

Set a per-key dollar ceiling — qlaud’s gateway enforces it. Hit $20, requests fail gracefully. No more $5K bills from a viral spike or a runaway loop.

Enforced by the gateway, not the model client.

How the gateway works

Multi-model is real

Claude one turn, DeepSeek the next

Same chat, swap models. Subagents on cheap models while the parent stays on the flagship. Provider failover when one’s down. Same Anthropic-shape API everywhere.

Claude, GPT, Gemini, Grok, Llama, DeepSeek, Qwen, Kimi, Mistral, Groq, Cerebras.

See routable models

Open source agent

MIT-licensed, fork it, audit it

qcode is open source. Read the agent loop. Run your own build. Self-host the desktop binary if your security team needs it. No vendor lock on the agent itself.

github.com/qlaudAI/qcode · MIT license.

Browse the source