OPALENGINE
ProjectsField NotesDocsGitHub ↗
Relay · Alpha
Get started
IntroductionQuickstartConcepts
Run a pool
GPU BackendsCLIConfigurationNodes & WindowsTesterSpeculative Decoding
Use it
Chat & ClientsAPIBenchmarks
Reference
SecurityTroubleshootingKnown Issues
Project
Releases & VersioningUpgradingSupportContributing
Docs

Documentation.

Relay · Alpha

Relay pools VRAM across teammates’ machines into one OpenAI-compatible endpoint. The host owns the models; nodes lend VRAM as replicas or pipeline-shard stages. No silent CPU fallback. Relay is alpha and unaccredited.

Start with the introduction →

Get started

  • Introduction
  • Quickstart
  • Concepts

Run a pool

  • GPU Backends
  • CLI
  • Configuration
  • Nodes & Windows
  • Tester
  • Speculative Decoding

Use it

  • Chat & Clients
  • API
  • Benchmarks

Reference

  • Security
  • Troubleshooting
  • Known Issues

Project

  • Releases & Versioning
  • Upgrading
  • Support
  • Contributing
[email protected]
RSSGITHUBPRIVACY