AI Providers

AI Providers

release-notifier supports three AI providers. All use the same prompt and produce the same structured output format.

Comparison

Provider	Cost	Quality	Speed	Requires
Claude Sonnet	~$3/1M tokens	Best	Fast	Anthropic API key
GPT-4o	~$2.50/1M tokens	Best	Medium	OpenAI API key
GPT-4o-mini	~$0.15/1M tokens	Good	Fast	OpenAI API key
Ollama mistral	Free	Good	Slow (local)	Ollama instance
Ollama llama3.2	Free	Good	Fast (local)	Ollama instance

For most homelab use cases, gpt-4o-mini offers the best cost/quality balance. Claude Sonnet produces the most reliable structured output and urgency classification.

---

Claude (default)

Get an API key at console.anthropic.com.

# .env
ANTHROPIC_API_KEY=sk-ant-...
# AI_PROVIDER=claude  # optional, claude is the default

Claude uses the claude-sonnet-4-6 model. This is currently the recommended Anthropic model for everyday tasks.

---

OpenAI

Get an API key at platform.openai.com/api-keys.

# .env
AI_PROVIDER=openai
OPENAI_API_KEY=sk-...
OPENAI_MODEL=gpt-4o-mini   # or gpt-4o

gpt-4o-mini is recommended — it follows the structured prompt reliably and costs ~20x less than Claude Sonnet for the same task.

---

Ollama (free, local)

Run summaries completely locally with no API costs. Requires an Ollama instance running on your network.

Install and start Ollama

# Install
curl -fsSL https://ollama.com/install.sh | sh

# Pull a model
ollama pull mistral

# Start — bind to all interfaces so Docker containers can reach it
OLLAMA_HOST=0.0.0.0 ollama serve

If your firewall blocks the port:

sudo firewall-cmd --add-port=11434/tcp --permanent
sudo firewall-cmd --reload

Configure release-notifier

# .env
AI_PROVIDER=ollama
OLLAMA_URL=http://192.168.1.10:11434   # IP of the machine running Ollama
OLLAMA_MODEL=mistral

Recommended models

Model	Size	Notes
mistral	~4GB	Best quality for structured output
llama3.2	~2GB	Smaller and faster, slightly lower quality
llama3.1:8b	~5GB	Best quality but slowest

Note: Ollama responses take 10–60 seconds per summary depending on your hardware. Urgency classification may be less accurate with smaller models compared to Claude or GPT-4o.

Running Ollama as a Docker container

services:
  ollama:
    image: ollama/ollama
    container_name: ollama
    volumes:
      - ./ollama:/root/.ollama

  release-notifier:
    image: 57faruk57/release-notifier:latest
    environment:
      - AI_PROVIDER=ollama
      - OLLAMA_URL=http://ollama:11434
      - OLLAMA_MODEL=mistral

After starting, pull the model:

docker exec ollama ollama pull mistral

---

Switching providers

No rebuild is needed — just update .env and restart:

docker compose up -d
# or
docker restart release-notifier

Verify the active provider:

curl http://localhost:8080/api/health
# "ai_provider": "Claude"