DEV Community

novatechnolab
novatechnolab

Posted on

🚀 ide_model_ext — One Extension to Chat with GPT, Gemini, and Claude

#python #opensource #openai #ai

The Problem

If you're working with AI models, you've probably dealt with this:

  • Install openai for GPT
  • Install google-generativeai for Gemini
  • Install anthropic for Claude
  • Learn three different APIs
  • Manage three sets of authentication
  • Write adapter code every time you want to switch models

What if one package handled all of it?

Introducing ide_model_ext

ide_model_ext gives you a single, universal interface to interact with OpenAI (GPT-4o), Google Gemini, and Anthropic Claude — from Python scripts, the terminal CLI, or Jupyter notebooks.

Install 

pip install ide_model_ext
Enter fullscreen mode Exit fullscreen mode

Use in 3 Lines 

from ide_model_ext import ModelClient

client = ModelClient()  # auto-selects best available model
response = client.chat("Explain quantum computing")
print(f"{response.content}\n— {response.model} ({response.provider})")
Enter fullscreen mode Exit fullscreen mode

That's it. The client auto-detects which API keys you've set and picks the best available model. Every response tells you exactly which model and provider answered.

✨ Features

Feature Description
🔗 Unified API One ModelClient for GPT-4o, Gemini, Claude Sonnet, and more
🔀 Dynamic Selection Auto-picks the best model from your configured API keys
🔍 Transparent Routing Every response includes .model and .provider
🌊 Streaming Real-time streamed responses from all providers
📓 Jupyter Magics %ai and %%ai magic commands
💻 CLI Tool Chat, interactive REPL, model listing
🔌 Extensible Add providers by subclassing BaseProvider

Quick Setup

Set API keys for the providers you want (you only need one):

export OPENAI_API_KEY="sk-..."
export GOOGLE_API_KEY="AIza..."
export ANTHROPIC_API_KEY="sk-ant-..."
Enter fullscreen mode Exit fullscreen mode

Python Usage

Explicit Model 

from ide_model_ext import ModelClient

# Pick a specific model
client = ModelClient(model="claude-3-5-sonnet-20241022")
response = client.chat("Write a haiku about Python")
print(response.content)
Enter fullscreen mode Exit fullscreen mode

Switch Models On-the-Fly 

client = ModelClient(model="gpt-4o")
response1 = client.chat("Hello from GPT!")

client.set_model("gemini-2.0-flash")
response2 = client.chat("Hello from Gemini!")

print(response1.provider)  # "openai"
print(response2.provider)  # "gemini"
Enter fullscreen mode Exit fullscreen mode

Streaming 

for chunk in client.stream("Tell me a story"):
    print(chunk, end="", flush=True)
Enter fullscreen mode Exit fullscreen mode

CLI Usage 

# One-shot
ide-model-ext chat "What is AI?"

# Specific model
ide-model-ext chat --model gemini-pro "Summarize this"

# Interactive REPL
ide-model-ext interactive

# List all models
ide-model-ext models
Enter fullscreen mode Exit fullscreen mode

Jupyter Notebook 

%load_ext ide_model_ext

%ai Tell me a joke about programming

%ai_model gemini-2.0-flash

%%ai
Write a Python function that calculates
the Fibonacci sequence recursively.
Enter fullscreen mode Exit fullscreen mode

Works in Jupyter Notebook, JupyterLab, Google Colab, and VS Code notebooks.

Architecture 

Your Code / CLI / Jupyter
        │
   ModelClient          ← simple user-facing API
        │
   ModelRouter          ← resolves model → provider
      ┌──┼──┐
  OpenAI Gemini Claude  ← provider adapters
      └──┼──┘
   BaseProvider         ← abstract interface
Enter fullscreen mode Exit fullscreen mode

The key design decision: every response includes .model and .provider so you always know which model answered — even under dynamic routing.

Supported Models

Provider Models
OpenAI gpt-4o, gpt-4o-mini, gpt-4-turbo, gpt-4, gpt-3.5-turbo, o1-preview, o1-mini
Gemini gemini-2.0-flash, gemini-1.5-pro, gemini-1.5-flash, gemini-1.0-pro
Claude claude-3-5-sonnet, claude-3-5-haiku, claude-3-opus, claude-3-sonnet, claude-3-haiku

Links

Built by Novatechnolab

MIT Licensed • Contributions welcome!

If this saves you time, give it a ⭐ on GitHub!

Top comments (0)