Open-Weight Ecosystem & Model Hubs
16 companies
0 of 0 categories open
Relationship:
Loading the map…
On this page
Where open models are distributed and run: hosting hubs, local runtimes, inference engines, and the research groups and communities that publish weights and datasets. This is the supporting infrastructure around open-weight models. The labs that train them have their own category.
What to compare
Keep distribution and execution separate in your head. A hub and a local runtime solve different problems and often get used together. For runtimes, compare quantisation support, hardware requirements, throughput, and whether one set of tooling covers both a laptop and a server. Licence clarity varies more than it should.
16 companies in this category
Allen Institute for AI (OLMo)Nonprofit AI research institute, creator of the fully-open OLMo model family.
BigCodeOpen collaboration behind the StarCoder code-generation models.
CivitaiCommunity hub for sharing image-generation models and fine-tunes.
EleutherAINonprofit research collective behind early open-source GPT-style models.- GPT4All (Nomic AI)Nomic AI's tool for running open LLMs privately on a local machine.
Hugging FaceLargest hub for open-weight models, datasets, and ML libraries like Transformers.
LAIONNonprofit building large open datasets used to train open AI models.
LM StudioDesktop app for discovering and running local LLMs with a GUI.- MLC AIMachine learning compilation project enabling LLMs to run on diverse hardware.
OllamaPopular tool for running open-weight LLMs locally with a simple CLI.- OpenBMBOpen-source big model research group behind the MiniCPM model family.
RedPajama (Together)Open dataset and model reproduction project led by Together AI.- TheBlokeProlific community contributor known for quantized open-weight model releases.
UnslothLibrary and tooling for fast, memory-efficient LLM fine-tuning.- llama.cpp / ggmlHigh-performance C/C++ inference engine that popularized local LLM quantization.
vLLM ProjectHigh-throughput open-source inference and serving engine for LLMs.