An intelligent AI code companion for your terminal and VS Code, coupled with a high-performance completions API, managed virtual keys, and real-time token tracking.
No dependencies. One command. Run sleepy to start.
Why Sleepy
An AI coding companion that lives in your terminal. No IDE lock-in, no web editor — just you and your agent in the shell.
One API, many models. OmniRoute picks the best model per request with automatic failover when a provider is down.
Virtual keys with role-based limits. Generate, rotate, and revoke keys from the dashboard without touching your gateway config.
Token counts, cost, and request windows updated live. Know exactly what you spend, per window, per model.
Pay-as-you-go credits with per-window rate limits that reset on plan change. No surprise invoices, no hidden fees.
Invite friends with your link and earn bonus tokens and credit when they verify their account with Telegram.
No config files to hand-edit. No vendor accounts to juggle. Just a terminal, a key, and a gateway that handles the routing for you.
That's it.
01
curl the installer, or npm i -g @sleepy-ai/cli. No dependencies, no IDE lock-in. You're authenticated in seconds.
02
Create keys from the dashboard with per-window limits. Rotate or revoke them anytime without touching your gateway.
03
OmniRoute picks the best model per request and fails over automatically. Watch tokens and cost stream in real time.
Install the CLI or the VS Code extension, create your key, and route your first request through SleepyAI.
Everything you need to know about Sleepy. Can't find what you're looking for? Reach out to our support team.
Sleepy is a terminal-native AI coding companion coupled with a high-performance completions API. The CLI runs your agent in the shell, and OmniRoute proxies requests across multiple model providers with automatic failover.
One command: `curl -fsSL https://www.sleepyai.org/install | bash` on macOS/Linux, or `irm https://www.sleepyai.org/install.ps1 | iex` from PowerShell on Windows. You can also install globally with npm via `npm i -g @sleepy-ai/cli`. No other dependencies.
No. The CLI uses authenticated sessions, so you can start with the free tier immediately. API keys are available on paid plans for direct API access and are managed from the dashboard.
Plans come with a monthly usage allowance. Extra usage is metered as credits you can top up anytime. Per-window rate limits (5h / 24h / weekly) protect you from surprise bills, and usage resets when you change plans.
SleepyAI routes across multiple providers, so you get access to many models through a single API. Browse the full list at /models, and the router picks the best model per request with failover.
SleepyAI doesn't store any of your prompt or completion data, rather it just proxies the API call from original provider. The original providers may use your anonymous prompt and completion data for AI training.