Think a little deeper.
Exploring how focused training helps models understand Python problems, plan changes, and check their work.
AN INDEPENDENT AI LAB
We train open-source models for focused expertise. Starting with Python, we’re working toward useful AI on modest hardware.

01OUR FOCUS
We build for developers, researchers, and independent builders who want open models they can adapt and run themselves. Our focus is specialized capability with lower hardware demands.
Exploring how focused training helps models understand Python problems, plan changes, and check their work.
Our first project specializes Gemma 4 for Python development: navigating repositories, fixing code, and using tests.
Working toward models for low-end hardware, with memory use, speed, and task quality measured together.
02THE WAY WE WORK
Our work starts with an open foundation and a question: what could this model become?
Inside our researchSelect open models and data suited to the task, with attention to their licenses and limitations.
Shape capabilities through focused fine-tuning, experimentation, and careful iteration.
Test what improves, understand what doesn’t, and use those findings to guide the next experiment.
03FROM THE LAB
Our first project is a Python specialist built on Gemma 4. We’re developing and testing it through Google’s Gemma 4 Developer Agent Competition on Kaggle, with an open-source release as the goal.
Explore the Python specialistWHY FORTUNA?
Our name comes from Fortuna, the goddess of fortune. For us, the opportunity to build useful intelligence is a fortunate beginning. Helping others build with it is the point.
Meet the idea behind the labGOOD THINGS BEGIN WITH A CONVERSATION
Research partnerships, funding,
and a shared belief in open intelligence.