d_model is an AI research lab that partners with frontier labs to develop interpretability and alignment researchers, including RL environments for open-ended interpretability tasks.