Skip to editor and terminal

Model Coder

Run OpenAI-style Python in your browser.

Tab through sample selection, run controls, layout reset, and theme switch.

📃 Selecting a different sample loads its starter code in the editor.

PyScript loading...

Model initializing...

Editor

Not run yet

This is the code editor area. Use Tab to move to the pane splitter and terminal after editing.

Use Up and Down arrow keys to resize editor and terminal panes.

Terminal

Interactive print() and input()

This terminal shows script output and supports input prompts when code requests input.

About This App

Model Coder is an educational sandbox for learning basic OpenAI-style Python coding syntax. It lets you experiment with prompts, responses, streaming, and async patterns in a safe environment.

The app runs entirely in your browser using PyScript for Python execution and supports dual-mode inference:

GPU Mode (WebGPU): Uses Phi-3-mini-4k-instruct for faster inference on supported GPUs
CPU Mode (wllama): Uses SmolLM2-360M-Instruct for CPU-only environments

Toggle between modes using the GPU/CPU button in the toolbar. No prompts, code, or output are sent to external services.

Use the following parameters in your code to connect to the local inference engine:

base_url (the OpenAI-compatible endpoint): https://local/openai
api_key (the authentication key): key123
model (the model name): local-llm

Prompts are filtered locally using a moderation word list. If a blocked term is detected, the app returns: I'm sorry. I can't help with that..

To get started, select a sample from the dropdown or write your own code in the editor pane, then click the run button. Check the model status in the top left to see when the model is ready to use. Happy coding!