Predictive Auditing of Hidden Tokens in LLM APIs via Reasoning Length Estimation

Open in new window