Provider Setup Guide¶

Orion supports both local (self-hosted) and cloud AI providers. You can switch between them at any time through the Dashboard Settings page or the CLI.

Provider Categories¶

Orion uses four provider categories, each configurable independently:

Switching Providers via Dashboard¶

1. Open http://localhost:3001 and log in
2. Navigate to Settings from the sidebar
3. You will see four provider cards: LLM, Image, Video, and TTS

Each Card Shows¶

• Provider mode dropdown: Local (Ollama / ComfyUI) or Cloud (OpenAI / Replicate)
• Model dropdown: available models for the selected mode
• Connection status: green check (connected), red alert (disconnected), or spinner (checking)
• Test Connection button: click to verify the selected provider is reachable before saving
• Model Parameters accordion: expand to adjust generation settings (e.g., temperature, max tokens)

To Switch a Provider¶

1. Select the desired mode from the Provider dropdown
2. The Model dropdown updates to show only models available for that mode
3. Select a model
4. Click Save -- the configuration is applied immediately
5. Optionally click Test Connection to verify the provider is reachable
6. The status indicator will update to show whether the new provider is reachable

Example: Switch LLM from Local to Cloud¶

1. On the LLM (Text Generation) card, change Provider to Cloud (OpenAI / Replicate)
2. Select GPT-4o from the Model dropdown
3. Click Save
4. The status indicator should turn green if the OPENAI_API_KEY environment variable is configured

Switching Providers via CLI¶

# View current provider configuration
orion provider status

┌──────────────────┬──────────┬──────────────┬───────────┐
│ Service          │ Mode     │ Model        │ Status    │
├──────────────────┼──────────┼──────────────┼───────────┤
│ LLM              │ LOCAL    │ llama3.2     │ connected │
│ Image            │ LOCAL    │ sdxl         │ connected │
│ Video            │ LOCAL    │ animatediff  │ connected │
│ TTS              │ LOCAL    │ piper        │ connected │
└──────────────────┴──────────┴──────────────┴───────────┘

Switch to cloud providers:

# Switch LLM to OpenAI GPT-4o
orion provider switch llm --mode CLOUD --provider openai --model gpt-4o

# Switch image generation to DALL-E 3
orion provider switch image --mode CLOUD --provider openai --model dall-e-3

# Switch TTS to ElevenLabs
orion provider switch tts --mode CLOUD --provider elevenlabs --model elevenlabs

# Switch back to local
orion provider switch llm --mode LOCAL --provider ollama --model llama3.2

Environment Variables for Cloud Providers¶

Cloud providers require API keys. Add these to your .env file:

# OpenAI — for GPT-4o, GPT-4o Mini, DALL-E 3, OpenAI TTS
OPENAI_API_KEY=sk-...

# Replicate — for Flux, Stable Video Diffusion
REPLICATE_API_TOKEN=r8_...

# Fal.ai — for image and video generation
FAL_KEY=fal_...

# ElevenLabs — for voice synthesis
ELEVENLABS_API_KEY=el_...

# Runway — for video generation
RUNWAY_API_KEY=rw_...

After updating .env, restart the affected services:

docker compose -f deploy/docker-compose.yml restart media editor director

Local Provider Setup¶

# Pull a model (inside the container)
docker compose -f deploy/docker-compose.yml exec ollama ollama pull llama3.2

# Verify available models
curl http://localhost:11434/api/tags

docker compose -f deploy/docker-compose.yml --profile gpu up -d

curl http://localhost:8188/system_stats

When to Use Local vs Cloud¶

Next Steps¶

• Full Pipeline Demo -- End-to-end walkthrough
• Monitoring -- Track provider health in Grafana
• Configuration Reference -- All environment variables
• Analytics Guide -- Track costs by provider
• System Administration -- Service health and GPU monitoring