LLM agents built for control. Designed for real-world use. Deployed in minutes.

Democratizing Reinforcement Learning for LLMs

Make beautiful isometric infrastructure diagrams

Natural Gradient Boosting for Probabilistic Prediction

Nano vLLM

Kyutai's Speech-To-Text and Text-To-Speech models based on the Delayed Streams Modeling framework.

ThunderSVM: A Fast SVM Library on GPUs and CPUs

Dynamic Memory Management for Serving LLMs without PagedAttention

A simple tui to view & control docker containers

Query-agnostic KV cache eviction: 3–4× reduction in memory and 2× decrease in latency (Qwen3/2.5, Gemma3, LLaMA3)

Efficient and easy multi-instance LLM serving

Time-Series-Optimized Transformer for Observability

Open-source implementation of AlphaEvolve

cuDF - GPU DataFrame Library

A privacy-first, self-hosted, fully open source personal knowledge management software, written in typescript and golang.

Debug, evaluate, and monitor your LLM applications, RAG systems, and agentic workflows with comprehensive tracing, automated evaluations, and production-ready dashboards.

MiMo: Unlocking the Reasoning Potential of Language Model – From Pretraining to Posttraining

real time face swap and one-click video deepfake with only a single image

Official Repository of Absolute Zero Reasoner

Benchmark suite for LLMs from Fireworks.ai

LTX-Video Support for ComfyUI

📅 A sleek and highly customizable calendar card for Home Assistant, designed for performance and a clean, modern look.

Model swapping for llama.cpp (or any local OpenAPI compatible server)

Programmatic sandboxing tool

A Flexible Framework for Experiencing Cutting-edge LLM Inference Optimizations

Build Real-Time Knowledge Graphs for AI Agents

Official implementation of Half-Quadratic Quantization (HQQ)