Xinference xinference.co ↗

Xinference v2.10.0

Release Notes

Enterprise-grade authentication, observability enhancements, and a brand-new Web UI — empowering AI inference at scale.

Latest Release 📦 pip install 'xinference==2.10.0'

📦 Installation

Choose your preferred method:

pip install 'xinference==2.10.0' Docker Image →

Highlights

🔐

Enterprise Auth & Security

  • OIDC / Keycloak SSO
  • IP & API Key Banning
  • Rate Limiting
  • Audit Logging
  • API Key Description & Status
⚠️ Disabled by default in v2.10.0; will be default-enabled in v3.0.
📊

Observability

  • Per-model GPU memory metrics
  • Grafana GPU Dashboard
  • Security Audit Dashboard
  • Filebeat config examples
  • Structured JSON logging

⚠️ Grafana/Filebeat require self-deployment.

⚠️ JSON logs via XINFERENCE_LOG_FORMAT=json.

🎨

New Web UI

  • Redesigned model registration & launch
  • Cache management & env variables UI
  • Optimized menu structure
  • Updated sidebar, logo & favicon
  • Server-side launch history
Continuously evolving
Community

🌐 Community Edition

New features, enhancements, and fixes available to everyone.

New Features

📊 Observability Enhancements

🛠 Enhancements

🐞 Bug Fixes

📚 Documentation

Enterprise

🏢 Enterprise Edition

Stability, security, and reliability improvements for production deployments.

🛡

Stability & Security Fixes

Numerous stability and security patches for enterprise-grade reliability.

🔧

Enterprise Deployment

Improved reliability and maintainability in production environments.