Model as String - Definable AI

Instead of importing model classes, pass a string to Agent(model=...):

from definable.agent import Agent

# Full format: "provider/model-id"
agent = Agent(model="openai/gpt-4o")
agent = Agent(model="anthropic/claude-sonnet-4-20250514")
agent = Agent(model="google/gemini-2.0-flash-001")
agent = Agent(model="deepseek/deepseek-chat")

# Bare model name defaults to OpenAI
agent = Agent(model="gpt-4o")
agent = Agent(model="gpt-4o-mini")

Supported Providers

Provider	Format	Example
OpenAI	`openai/model-id`	`openai/gpt-4o`
Anthropic	`anthropic/model-id`	`anthropic/claude-sonnet-4-20250514`
Google Gemini	`google/model-id`	`google/gemini-2.0-flash-001`
DeepSeek	`deepseek/model-id`	`deepseek/deepseek-chat`
Mistral	`mistral/model-id`	`mistral/mistral-large-latest`
Moonshot	`moonshot/model-id`	`moonshot/moonshot-v1-8k`
xAI	`xai/model-id`	`xai/grok-3`
Perplexity	`perplexity/model-id`	`perplexity/sonar-pro`
Ollama	`ollama/model-id`	`ollama/llama3`
OpenRouter	`openrouter/model-id`	`openrouter/meta-llama/llama-3-70b`

When to Use Class Imports

Use the class-based approach when you need to configure model-specific options:

from definable.model.openai import OpenAIChat

model = OpenAIChat(
    id="gpt-4o",
    api_key="sk-...",          # Custom API key
    strict_output=False,       # Disable strict JSON mode
    base_url="https://...",    # Custom endpoint
)

String shorthand uses default configuration for each provider.