pre.dev

Hard, verified RL tasks for coding agents, built from private production code. Each ships in a runnable environment with a graded verifier. Hard, verified coding tasks. From code no model has seen. Long-horizon RL tasks for post-training. Tasks, not repo access: each ships in a runnable environment with a graded verifier, a hidden reference solution, and solver trajectories. Coordinate payments across service instances. August 2026 sweep. No model fully solves this task. Scale Pipeline and corpus, updated every minute Each task recreates one real commit from a licensed repo. A reference solution scores at least 0.95, so every task is solvable, and it stays hidden from the solver. Graded checks give continuous reward from 0 to 1, including partial progress.