H2O Open-Source Suite

Status der H2O-Komponenten

Komponente	Status	URL	Notiz
H2O Flow (AutoML)	live	h2oflow.ben-e-fit.ai	Web UI fuer AutoML-Training
h2oGPT	pending	—	3GB GPU-Image, GCR-Registry
H2O Wave	pending	—	Image privat, Alternative zu pruefen
H2O LLM Studio	pending	—	8GB GPU-Image
H2O Document AI	pending	—	Alternative paperqa pruefen

Verwende stattdessen

Bis die H2O-Komponenten verfuegbar sind, stehen aequivalente Open-Source-Tools bereit:

Use-Case	Empfehlung	URL
Open-Source LLM Web UI (h2oGPT-Aequivalent)	Open WebUI + Ollama	chat.ben-e-fit.ai
AutoML	H2O Flow	h2oflow.ben-e-fit.ai
RAG ueber Dokumente	OWUI Knowledge + bge-m3	chat.ben-e-fit.ai oder chat.medialine.app
Workflow-Automation	n8n	n8n.ben-e-fit.ai
Vector-Store (LLM Studio Aequivalent fuer RAG)	Qdrant + bge-m3	qdrant.ben-e-fit.ai

W31 Phase 2 Roadmap

Beim naechsten Image-Pull-Window:

• h2oGPT via gcr.io/vorvan/h2oai/h2ogpt-runtime:0.2.1 (3GB)
• H2O LLM Studio via gcr.io/vorvan/h2oai/h2o-llmstudio:latest (8GB)
• H2O Wave: Alternative pruefen (Image-Private)
• CF-Tunnel-Routes anlegen via Browser-Cookie-Auth