Founder Bench
Three frontier models each ran the same seed-stage software company: 13 cloned apps, a team, customers, and six months of real work.
A task passes only if every check is met and no rule is broken.
What was tested
Leaderboard
One task at a time
A whole company
What it costs, how fast it goes
Run your model on it
We will run any model you name through the same company and send you every transcript, tool call and grader verdict.
