Settings

Test catalog, methods, reference ranges, and AI configuration

AI Review Provider

Live

Provideropenai-compatible

Modelnvidia/Gemma-4-31B-IT-NVFP4

Base URLhttp://host.docker.internal:8001/v1

API key•••••••• set

Configure via environment variables

# Local Gemma (Ollama) — Endyne's chosen path
AI_PROVIDER=ollama
AI_BASE_URL=http://localhost:11434/v1
AI_MODEL=gemma3:4b

# Anthropic Claude (alternative)
# AI_PROVIDER=anthropic
# AI_API_KEY=sk-ant-...
# AI_MODEL=claude-haiku-4-5

# Any OpenAI-compatible (Groq, Together, OpenRouter, LM Studio, vLLM)
# AI_PROVIDER=openai-compatible
# AI_BASE_URL=https://api.groq.com/openai/v1
# AI_API_KEY=...
# AI_MODEL=llama-3.3-70b-versatile

When unset, the LIMS uses deterministic scripted output so the demo never blanks. Live AI activates the moment the env vars are present and the service is reachable.

Current: openai-compatible · nvidia/Gemma-4-31B-IT-NVFP4 Inference reachable

Lab catalog

Test catalog, methods, holding times, reference ranges, QC criteria, EDD format mapping — managed via Prisma schema + seed for v1. A full admin UI lands in v1.1.