Code Hub

High-throughput LLM inference and serving engine with PagedAttention.

Efficient LLM inference on CPU/edge; a de-facto standard for quantized inference.

Run open LLMs locally with one command; a go-to for local inference.

High-performance serving for complex LLM programs; structured generation & concurrency.

Hugging Face's production-grade LLM inference server.

General framework for building LLM apps and agents; widest ecosystem.

Visual LLM app platform for workflows and agents; self-host friendly.

Self-hostable workflow automation with AI nodes for orchestration.

Open-source software-engineering agent that reads/writes code and runs commands.

Data framework to connect private data to LLMs; a common RAG foundation.

Vector database for large-scale similarity search; common for RAG.

Unified fine-tuning framework supporting hundreds of models and efficient methods.

Faster, lower-VRAM LLM fine-tuning that runs on consumer GPUs.

Parameter-efficient fine-tuning (LoRA, etc.) for low-cost adaptation.

Node-based image/video generation workflows for diffusion pipelines.

OpenAI's general-purpose speech recognition; a multilingual transcription baseline.