<p>In daily conversations, people integrate multimodal cues from faces and adapt their gaze patterns to enhance comprehension according to the situation. Here we used eye-tracking to explore whether autistic individuals can adjust their face-viewing patterns flexibly when speech audio and visual information are asynchronous, with a primary focus on preschool-aged children. Participants from preschool cohort, including 72 autistic and 57 non-autistic individuals aged 3–6 years, watched videos of an actress speaking in either temporal synchrony or asynchrony conditions. We analyzed participants’ attention to the face, particularly the mouth, in both conditions. Full-sample data (ages 3–17 years, <i>N</i> = 228) are reported in supplementary materials for broader developmental context. We analyzed participants’ attention to the face, particularly the mouth, in both conditions. We discovered that both groups looked more at the face in asynchrony conditions and more at the mouth in synchrony conditions, suggesting they can detect speech audio-visual synchrony (AVS) and adjust their face-viewing patterns accordingly. Additionally, we found that the autistic group looked less at the face than the non-autistic group in both conditions. Our findings suggest that, like non-autistic individuals, autistic individuals are sensitive to speech AVS and can adjust their gaze patterns accordingly. However, these abilities may be hindered by their reduced face-looking time, resulting in less efficient and flexible integration of speech AVS, potentially impeding language development. These results provide novel insights for interventions in language development.</p>