Sprites
Sprites are persistent, hardware-isolated Linux environments designed for running arbitrary code with stateful checkpoint and restore capabilities. They function as dedicated microVMs that hibernate when idle and wake automatically, making them ideal for AI agents, development environments, and secure code execution.
winapp CLI v0.5.0
The Windows App Development CLI (winapp CLI) is a unified command-line interface designed to streamline the development of Windows applications. It simplifies complex tasks such as managing Windows SDKs, handling MSIX packaging, generating app identity, and configuring manifests and certificates for various app frameworks.
dynwinrt v0.1.0-preview.18
dynwinrt is a dynamic Windows Runtime (WinRT) API projection for Node.js and Electron that allows developers to call native Windows APIs directly from JavaScript or TypeScript. It eliminates the need for writing native C++ addons or complex build steps by reading .winmd metadata and invoking COM vtables at runtime via libffi.
DeepSeek-V4-Flash-0731 v0731
DeepSeek-V4-Flash-0731 is an official, re-post-trained version of the DeepSeek-V4-Flash model, featuring significantly enhanced agentic capabilities and a built-in DSpark speculative decoding module. It is designed for high-efficiency, large-scale tasks, offering competitive performance in coding, reasoning, and multi-step tool use.
ThunderAgent
ThunderAgent is a program-aware scheduling system designed to optimize high-throughput agentic inference. By treating agentic workflows as schedulable programs rather than independent requests, it effectively eliminates KV cache thrashing and improves multi-node scaling.
Marker v2.0.0
Marker is a high-performance document intelligence pipeline designed to convert PDFs, images, and various office documents into clean Markdown, JSON, HTML, or chunks. It leverages advanced deep learning models, including the Surya VLM, to handle complex layouts, tables, equations, and multi-language text extraction with high accuracy.
Scriptc v0.0.17
scriptc is a TypeScript-to-native compiler that transforms ordinary TypeScript code into small, fast, self-contained executables. It eliminates the need for Node.js, V8, or any embedded JavaScript engine for statically compiled code, resulting in significantly faster startup times and smaller binary sizes.
AngelSpec v0.1.0
AngelSpec is a unified, torch-native training framework designed for speculative decoding, supporting both autoregressive Multi-Token Prediction (MTP) and block-parallel drafting architectures. It enables independent scaling of inference and training through a disaggregated architecture, facilitating high-performance model deployment.
Qwythos-27B-v1 v1
Qwythos-27B-v1 is an open-weight, full-parameter reasoning model built on the Qwen3.5-27B base. It is designed for advanced technical tasks, featuring native multi-token prediction, a full vision tower, and a 1M-token context window.
AgentENV v0.1.0
AgentENV (AENV) is a distributed, open-source platform designed for running isolated AI agent environments at scale. It utilizes Firecracker microVMs to provide secure, kernel-level isolation while enabling rapid snapshotting, pausing, and forking of environments to support efficient agentic reinforcement learning workflows.
Codex Security v0.1.1
Codex Security is an application security agent designed to help development teams identify, validate, and remediate vulnerabilities within their codebases. It provides both a CLI and a TypeScript SDK to scan repositories, review changes, and propose actionable fixes for security issues.
LFM2.5-Encoder-350M v2.5
LFM2.5-Encoder-350M is a multilingual bidirectional encoder model built on the LFM2 hybrid architecture. It is designed for high-performance, on-device tasks such as text classification, retrieval, and semantic similarity across 15 languages.