Choose Your AI Model

Supported Providers

Provider	Models	Key Command	Status
Anthropic	Claude Sonnet 4.6 (default), any Claude model	`vault set anthropic_api_key`	`ready`
Ollama (local)	Llama 3, Mistral, Qwen, any local model	(no key needed)	`ready`
Ollama Cloud	Hosted Ollama models (validated end-to-end with `glm-5.1`)	`vault set openai_compat_api_key`	`ready`
OpenAI	GPT-4o + Azure OpenAI	`vault set openai_api_key`	`beta`
Google	Gemini 2.0 Flash	`vault set gemini_api_key`	`beta`
xAI	Grok models	`vault set xai_api_key`	`beta`
Any OpenAI-compatible	LM Studio, vLLM, Groq, DeepSeek, custom endpoints	`vault set openai_compat_api_key`	`beta`

ready means validated end-to-end in production use. beta means the adapter is code-complete and passes tests, but still needs real-world validation — test reports are very welcome.

Quick Setup

# Cloud providers (pick one)
aibutler vault set anthropic_api_key sk-ant-...
aibutler vault set openai_api_key sk-...
aibutler vault set gemini_api_key AIza...

# Local (no key needed — just install Ollama)
# https://ollama.com
ollama pull llama3

For a local model, point settings.model at it in ~/.aibutler/config.yaml:

settings:
  model: llama3    # any non-cloud model name routes to your local endpoint

For Ollama Cloud, set the endpoint, pick a hosted model, and store your key:

settings:
  model: glm-5.1              # any model hosted on Ollama Cloud

configurations:
  models:
    base_url: https://ollama.com

aibutler vault set openai_compat_api_key <your-ollama-cloud-key>

Switching Models

Change model at any time in ~/.aibutler/config.yaml:

settings:
  model: claude-sonnet-4-6    # or: gpt-4o, gemini-2.0-flash, llama3

The model name picks the provider automatically: claude* routes to Anthropic, gpt* to OpenAI, gemini* to Google, grok* to xAI, and anything else to your local OpenAI-compatible endpoint (Ollama by default).

Check the active model in the REPL:

/model

Cost Management

AI Butler tracks every token and dollar spent:

aibutler cost status      # Current spend
aibutler cost breakdown   # Per-model breakdown
aibutler cost budget 50   # Set $50/month budget

Strategies: frugal (minimize cost), balanced (default), quality (maximize quality).

Offline Mode

Run completely offline with local models:

settings:
  offline_mode: true
  model: llama3

No data leaves your machine. Zero cloud API calls.