A dataset of 53.6K real-world in-IDE code edits of AI-generated code in Python, TS, and JS.

A high-resolution corpus for cross-benchmark AI agent evaluation.

High-quality RAG framework with zero or minimal ingestion cost.

An open-source governed pipeline for directing AI models through structured security tasks to find and fix vulnerabilities.

A long-context multi-session medical dialogue benchmark for evaluating LLM reasoning over patient histories.

Free and self-hosted document signing service.

Super Mario 64 as a library for use in external game engines.

An OCI runtime that allows containers to run system services (like Docker) without needing --privileged mode.

A TypeScript-to-Native compiler that removes the need for a JS engine in the final binary.

An interactive simulation and visualization of PostgreSQL internal operations.

A network tool for building proxies and bypassing network censorship using protocols like VLESS and Reality.

A staff-engineer layer for AI coding agents to orchestrate complex software engineering tasks.

An extremely fast Python linter and formatter written in Rust

A WASM-4 runtime for Java ME phones

A complete voice model with only 9.36M parameters for efficient voice synthesis.

A CLI email client written in Rust for managing emails via terminal.

Python library for an open-weights background removal model

Codebase containing emerging algorithms for LLM optimization including SOAP and Muon

A fully typed functional programming library for TypeScript, providing a correctness-focused approach to app development.

A distributed blockwise attention mechanism for efficient long-sequence LLM inference.