Providers

Calliope CLI supports 12+ AI providers. Switch between them instantly without changing your workflow.

Supported Providers

Cloud Providers

Provider	Command	Default Model	API Key Variable
Anthropic	`/provider anthropic`	claude-sonnet-4-20250514	`ANTHROPIC_API_KEY`
OpenAI	`/provider openai`	gpt-4o	`OPENAI_API_KEY`
Google	`/provider google`	gemini-2.0-flash	`GOOGLE_API_KEY`
Mistral	`/provider mistral`	mistral-large-latest	`MISTRAL_API_KEY`
Groq	`/provider groq`	llama-3.3-70b-versatile	`GROQ_API_KEY`
Together	`/provider together`	meta-llama/Llama-3.3-70B-Instruct-Turbo	`TOGETHER_API_KEY`
Fireworks	`/provider fireworks`	accounts/fireworks/models/llama-v3p3-70b-instruct	`FIREWORKS_API_KEY`
AI21	`/provider ai21`	jamba-1.5-large	`AI21_API_KEY`
HuggingFace	`/provider huggingface`	meta-llama/Llama-3.3-70B-Instruct	`HUGGINGFACE_API_KEY`
DeepSeek	`/provider deepseek`	deepseek-chat	`DEEPSEEK_API_KEY`

Gateway Providers

Provider	Command	Description	API Key Variable
OpenRouter	`/provider openrouter`	Access any model via unified API	`OPENROUTER_API_KEY`
LiteLLM	`/provider litellm`	Self-hosted proxy to multiple providers	`LITELLM_BASE_URL`, `LITELLM_API_KEY`

Local Providers

Provider	Command	Description	Configuration
Ollama	`/provider ollama`	Run models locally	`OLLAMA_BASE_URL` (default: localhost:11434)

Switching Providers

During a Session

Switch providers at any time:

calliope> /provider google
Provider set to: google

calliope> /provider anthropic
Provider set to: anthropic

View Available Providers

See which providers are configured:

calliope> /provider
Current: anthropic
Available: anthropic, google, openai, mistral

Only providers with valid API keys are shown.

Auto Selection

Set provider to auto to let Calliope choose the best available:

calliope> /provider auto

Priority order: Anthropic > OpenAI > Google > Mistral > OpenRouter > Together > Groq > Ollama > LiteLLM

Configuring Models

Set a Specific Model

Override the default model for your current provider:

calliope> /model claude-opus-4-20250514
Model set to: claude-opus-4-20250514

View Current Model

calliope> /model
Model: claude-sonnet-4-20250514

Model Availability by Provider

Model names vary by provider. Consult each provider’s documentation for available models.

Provider Details

Anthropic (Claude)

Best-in-class for coding tasks and tool use.

export ANTHROPIC_API_KEY=sk-ant-...

Available models:

claude-sonnet-4-20250514 (default) - Best balance of speed and capability
claude-opus-4-20250514 - Most capable
claude-3-5-haiku-20241022 - Fastest

Get an API key: console.anthropic.com

OpenAI (GPT)

Industry standard with excellent general capabilities.

export OPENAI_API_KEY=sk-...

Available models:

gpt-4o (default) - Multimodal flagship
gpt-4-turbo - Previous generation flagship
gpt-4o-mini - Smaller, faster

Get an API key: platform.openai.com

Google (Gemini)

Google’s flagship AI models.

export GOOGLE_API_KEY=...

Available models:

gemini-2.0-flash (default) - Fast and capable
gemini-1.5-pro - Larger context window
gemini-1.5-flash - Balanced

Get an API key: aistudio.google.com

Mistral

European AI company with strong open-weight models.

export MISTRAL_API_KEY=...

Available models:

mistral-large-latest (default) - Most capable
codestral-latest - Optimized for code
mistral-medium-latest - Balanced

Get an API key: console.mistral.ai

Groq

Ultra-fast inference on open models.

export GROQ_API_KEY=...

Available models:

llama-3.3-70b-versatile (default) - Llama 3.3
mixtral-8x7b-32768 - Mixtral

Get an API key: console.groq.com

Together

Platform for open-source models.

export TOGETHER_API_KEY=...

Available models:

meta-llama/Llama-3.3-70B-Instruct-Turbo (default)
Many open models available

Get an API key: api.together.xyz

OpenRouter

Unified API to access any model from any provider.

export OPENROUTER_API_KEY=...

Model format: provider/model-name

anthropic/claude-sonnet-4 (default)
openai/gpt-4o
google/gemini-2.0-flash

Get an API key: openrouter.ai

Ollama (Local)

Run models locally on your machine.

# Install Ollama first: https://ollama.ai
export OLLAMA_BASE_URL=http://localhost:11434

Pull a model:

ollama pull llama3.3

Available models: Any model you pull with ollama pull

llama3.3 (default)
codellama
mistral
Many more at ollama.ai/library

DeepSeek

Chinese AI company known for strong coding and reasoning capabilities.

export DEEPSEEK_API_KEY=...

Available models:

deepseek-chat (default) - General chat model
deepseek-coder - Optimized for coding tasks
deepseek-reasoner - Enhanced reasoning

Get an API key: platform.deepseek.com

LiteLLM (Proxy)

Self-hosted proxy that provides a unified interface to multiple providers.

export LITELLM_BASE_URL=http://localhost:4000
export LITELLM_API_KEY=...  # Optional

Setup: See LiteLLM documentation

Provider Priority

When using auto provider or when your preferred provider is unavailable:

Anthropic - Preferred for coding tasks
OpenAI - Strong general capabilities
Google - Good multimodal support
Mistral - European alternative
OpenRouter - Access to many models
Together - Open-source models
Groq - Fast inference
Ollama - Local fallback
LiteLLM - Proxy fallback

Tips

Use Different Providers for Different Tasks

# Complex reasoning
calliope> /provider anthropic
calliope> Help me design a microservices architecture

# Quick questions
calliope> /provider groq
calliope> What's the syntax for a Python list comprehension?

# Local/offline work
calliope> /provider ollama
calliope> Review this code

Cost Optimization

Use Groq or Together for simple tasks (often cheaper)
Use Ollama for development/testing (free)
Reserve Anthropic/OpenAI for complex tasks

Fallback Strategy

Configure multiple providers so you have fallbacks:

export ANTHROPIC_API_KEY=...  # Primary
export OPENAI_API_KEY=...     # Backup
export OLLAMA_BASE_URL=...    # Offline fallback

Bring Your Own Keys Autonomous Loops