AI at the center
Model training and AI research are the purpose of the lab. We start with the capabilities we want to understand and improve.
THE IDEA BEHIND THE LAB
FortunaML is an independent, AI-first startup and research lab focused on training open models.
We build for people who want open, trained models they can adapt and use themselves. Our first project specializes Gemma 4 for Python, with lower hardware requirements as a continuing research priority.

01OUR NAME
Fortuna is the goddess of fortune. We chose her as a symbol of possibility—and a reminder to do something meaningful with it.
We feel fortunate to work on models that could help people think, create, and solve difficult problems. We want that good fortune to travel further: to the researchers, developers, and curious people who build with what we make.
The name is a starting point. The work is what gives it meaning.
02BEHIND FORTUNAML
You can find me on GitHub as BigDecimals.
I’m building FortunaML around a simple ambition: train useful specialist models, share the work openly, and make progress toward AI that more people can run on the hardware they have.
The Gemma 4 Python specialist is our first project. I’m seeking funding to continue the research, improve our training and evaluation, and turn experiments into documented open releases.
Discuss funding or research support03WHAT GUIDES US
We’re early in our journey. These are the ideas we want to keep close as the lab grows.
Model training and AI research are the purpose of the lab. We start with the capabilities we want to understand and improve.
We intend to share our models, code, and research openly, with permissive licenses and clear documentation so others can build on the work.
Data choices, training decisions, and evaluation all shape a model. We believe each deserves thoughtful attention.
Interesting questions matter. So does connecting the answers to problems people actually want to solve.
GOOD THINGS BEGIN WITH A CONVERSATION
Research partnerships, funding,
and a shared belief in open intelligence.