The open AI community, built to move.
Build, discover, and collaborate on models, datasets, and AI applications in one open home.

Trending on this week
Curated paths into the public Hub. No feed request needed.Spaces
Browse AI appsDatasets
The home of open AI
Create, discover, and collaborate on ML, in one place.
The collaboration platform
Host and collaborate on unlimited public models, datasets, and applications.
Move faster
Build with the open source stack and commands already documented by MEGA.
Explore every modality
Search real public datasets across text, image, audio, and multimodal work.
Build your portfolio
Share your work with the world and make every release inspectable.
Accelerate your AI
Run open releases with production-grade platform tools.
Team & Enterprise
Move from open experiment to serious production.
Private releases, resource groups, audit trails, single sign-on, and direct support in one governed workspace.Explore Enterprise →Inference providers
Run models behind one familiar API, with routing and usage controls built in.
Spaces compute
Attach the right hardware to a live app, an endpoint, or a reusable Job.
Built in the open
MEGA grows alongside the formats, protocols, and tools the community already trusts.
megatensors
Python SDKDownload, inspect, and publish MEGA artifacts from Python.
MEGA Skills
Agent workflowsReusable workflows for builders and coding agents.
Hub Docs
DocumentationThe public guide to repositories, runtime, and collaboration.
Codex Plugin
Agent integrationMEGA capabilities packaged for agentic development.
Transformers
Compatible ecosystemFamiliar model loading paths and repository metadata.
Safetensors
Compatible formatSafe, inspectable model weight distribution.
Git LFS
Transport protocolVersioned large files through a workflow developers know.
Xet
Data planeChunk-aware transfer for large, repeatedly changing artifacts.
MCP
Tool protocolA standard bridge from agents into Hub knowledge and actions.
Gradio
Space interfaceTurn model work into a usable community application.
vLLM
Inference engineHigh-throughput serving paths for open model releases.
OpenDAL
Storage ecosystemComposable storage ideas across cloud and local backends.

