Easy AI

Working with artificial intelligence models often requires managing multiple vendor-specific libraries, API keys, and initialization parameters. Whether you're connecting to cloud providers like OpenAI or running models locally with Ollama, easy_ai provides a beginner-friendly helper function that unifies model creation across different AI providers.

A small real-world example

Imagine you're building an application that needs to interact with AI models. Depending on your environment, you might want to use a cloud provider like OpenAI for production or a local model with Ollama for offline testing, without rewriting your model initialization logic.

from py_simple import get_model

# Initialize a cloud model using OpenAI
openai_model = get_model(provider="openai", model_name="gpt-4o", api_key="your-api-key")

# Initialize a local model using Ollama
ollama_model = get_model(provider="ollama", model_name="llama3")

# Generate a response using the initialized model
response = ollama_model.invoke("Hello, world!")
print(response.content)

Example output:

Hello! How can I assist you today?

What happened?

get_model() provides a single, unified function to instantiate chat models from popular AI providers including OpenAI, Ollama, Anthropic, Google, and Mistral.

When provider="openai" is passed, get_model() imports and returns a ChatOpenAI instance configured with the specified model name and API key.

When provider="ollama" is passed, get_model() connects to your local Ollama server (defaulting to http://localhost:11434) and returns a ChatOllama instance ready to execute prompts locally.

You can also pass provider="anthropic", provider="google", or provider="mistral" to initialize ChatAnthropic, ChatGoogleGenerativeAI, or ChatMistralAI models with optional parameters like api_key, base_url, and timeout.

Why use these helpers?

Instead of installing and importing separate provider libraries (langchain_openai, langchain_ollama, langchain_anthropic, etc.) and writing different initialization code for each platform, you can simply write:

model = get_model(provider="openai", model_name="gpt-4o", api_key="your-api-key")

This keeps AI model setup simple, readable, and beginner-friendly while allowing you to easily switch between cloud and local AI providers with a single function call.