Benchmark It Yourself

Build a small personal eval harness on your own repo, because the public leaderboard is answering a different question than the one you're asking.

Part of: Coding on Open Models