Reliable deployment of large language models (LLMs) requires accurate uncertainty estimation. Existing answer-first methods produce confidence only after generating an answer, thereby measuring the correctness of a specific response and limiting their practical usability. We study a confidence-first paradigm, where the model outputs its confidence before answering, interpreting this score as the model’s probability of answering the question correctly under its current policy.

We propose CoCA(Co-optimized Confidence and Answers), a GRPO reinforcement learning framework that jointly optimizes confidence calibration and answer accuracy via segmented credit assignment. By assigning separate rewards and group-relative advantages to confidence and answer segments, CoCA enables stable joint optimization and substantially mitigates the observed reward-hacking failure mode. Experiments across math, code, and factual QA benchmarks show improved calibration and uncertainty discrimination while preserving answer quality, thereby enabling a broader range of downstream applications.