Models

Lab supports multiple AI providers and models. This reference covers available providers, popular models, and configuration.

Supported Providers

Built-in Shortcuts

For convenience, these shortcuts are available without the provider prefix:

Popular Models

OpenAI

Anthropic

Google

AWS Bedrock

Configuration

Environment Variables

Set API keys as environment variables:

# OpenAI
export OPENAI_API_KEY=sk-...

export ANTHROPIC_API_KEY=sk-ant-...

export GOOGLE_API_KEY=...

AWS Bedrock

Bedrock uses AWS credentials. Configure via:

• Environment variables (AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY)
• AWS credentials file
• IAM role (when running in AWS)

Default Model

Set a default model so you don’t need to specify it each time:

%config AiMagics.default_model_name = "anthropic:claude-3-5-sonnet-latest"

Listing Available Models

See what models are available:

# All models
%ai list

%ai list openai
%ai list anthropic

Creating Aliases

Register custom shortcuts:

# Create aliases
%ai register fast openai:gpt-4o-mini
%ai register best anthropic:claude-3-opus-latest

%%ai fast
Quick question about Python syntax

Choosing a Model

For Speed

Use smaller, faster models:

• gpt4o-mini
• claude-haiku
• gemini-2.0-flash

For Quality

Use larger, more capable models:

• gpt-4o
• claude (sonnet)
• claude-3-opus-latest

For Cost

Consider token usage and pricing:

• Haiku and mini models are most economical
• Use larger models only when needed
• Shorter prompts reduce costs

For Privacy

If data sensitivity is a concern:

• AWS Bedrock keeps data in your AWS account
• Check provider data policies