Difficulties engaging mental effort during cognitive tasks are well documented in individuals with ADHD and are reflected in the DSM-5-TR inattention criteria, which note a tendency to avoid or dislike tasks requiring sustained effort. The Diminishing Criterion Model (DCM) is a metacognitive framework of effort regulation that integrates objective performance and subjective confidence ratings to predict time investment. Although validated in neurotypical samples, the model has not been applied to neurodevelopmental populations. The present study applied DCM principles to an adult ADHD group (n = 70), a sub-threshold ADHD group (n = 72), and a control group (n = 71) recruited online through Prolific Academic. Groups were formulated using quota sampling to ensure balance across key demographic variables (e.g., age, gender, and education). Participants completed a 40-item word-analogy task that recorded item-by-item response times and confidence ratings. No group differences were found in overall accuracy, mean confidence, resolution, or calibration. However, control participants invested more time overall than the ADHD group. Effort-regulation analyses based on the DCM revealed distinct time-investment patterns: the ADHD group invested the least amount of time regardless of item difficulty. In contrast, the sub-threshold group resembled controls on easier and intermediate items but invested less time on the most challenging ones. These results expand on the limited research on metacognition in ADHD, which has primarily relied on one-shot measures of monitoring accuracy. The present findings provide preliminary evidence that effort is dynamically allocated differently during tasks in neurodevelopmental populations.