Platform · ls-judge
Run untrusted code in 65+ languages — each submission in its own microVM. The platform every code-running product of ours is built on.
How a submission flows
Products
Scheduler
ls-judge
HTTP API · WebSocket · webhooks
Hot lane
Warm snapshot pool · milliseconds
Cold lane
Snapshot restore on demand
Queue
Redis streams · stateless workers
Results stream back over WebSocket
Firecracker microVMs
Engineering
Security, speed and correctness are designed in — not bolted on.
Each submission boots in its own Firecracker microVM with a private network namespace, seccomp and a jailer — never a shared container.
Warm snapshot pools serve popular languages in milliseconds; everything else restores from a cold snapshot on demand.
Verdicts, run time and memory for every test case stream back over WebSocket as they finish — no polling.
Multi-file submissions, batch runs and webhooks behind a clean HTTP API, with real compiler and interpreter versions.
Redis-backed streams and stateless workers with token-based identity, so capacity scales horizontally.
A full-fleet stress and correctness gate exercises every runtime through the real production path before release.
Built on ls-judge
One judge, many front doors — on Android, in the browser and through the public API.
A multi-file IDE that opens in a browser tab.
Write in 80+ languages on the same engine that powers VS Code. Preview Markdown, diff files, test APIs and run snippets — with your files kept on your device, not our servers.
Compile and judge code in 61 languages.
Run code against test cases on your phone with real verdicts — accepted, wrong answer, time limit and more. Every run executes on ls-judge in its own isolated microVM, across 114 language runtimes.
Explore
FAQ
Yes. The playground needs no sign-up, and the free tier includes 500 executions / month · 60 requests / minute · playground with no sign-up. Higher-volume plans for products and teams are listed on the pricing page.
Every submission runs in its own Firecracker microVM — hardware virtualisation with a private network namespace, seccomp filters and a jailer — rather than a container shared with other users' code.
65+ languages across 120+ runtimes, including Python, C++, Java, JavaScript, TypeScript, Go, Rust, C#, Kotlin and many more, using real compiler and interpreter versions.
Per-test results — verdict, run time and memory — stream back over WebSocket as each test finishes. Verdicts include AC, WA, TLE, MLE, RE, CE.
Yes — that's exactly how Omni Compiler and the Online Code Editor use it. See the docs for the HTTP API, batch runs and webhooks, and get in touch for higher limits.
Start on the free tier today, or tell us about your volume and we'll help you find the right plan.