Ell

Ell treats prompts as programs, not strings: you write ordinary Python functions decorated with @ell.simple(model=...), and Ell handles turning them into LLM calls. Every prompt version is automatically serialized with autogenerated commit messages, so prompt iteration works like a training loop with checkpoints. Ell Studio is a local, open-source companion app for visualizing prompt versions, monitoring outputs, and catching regressions. Multimodal inputs (PIL images, audio) are first-class via type coercion.

```bash

pip install "ell-ai[all]"

```

```python

import ell

@ell.simple(model="gpt-4o")

def hello(world: str):

"""You are a helpful assistant.""" # system message

return f"Say hello to {world}." # user message

hello("world")

```

Launch the visualizer with ell-studio --storage ./logdir. Set your provider API key (e.g. OPENAI_API_KEY). Docs: docs.ell.so.