Heterogeneity among individuals on the autism spectrum is widely acknowledged as a barrier to develop effective interventions. Overcoming this challenge requires characterization of individual differences, especially for children that are minimally verbal and often excluded from research studies. Most studies that describe autistic subgroups identify a single minimally verbal verbal group based on a single identifying measure (e.g., ADOS module one or single item indicating absence of phrase speech). Determining personalized courses of intervention requires a more detailed understanding since a single intervention will not be effective for all who are minimally verbal. The present study identified comprehensive profiles of cognitive, language, and social communication skills within a large, diverse, group of minimally verbal children with autism. The analysis combined baseline data from two studies to yield a sample of 344 participants, who were 3 to 8 years old at the time of study onset, with 60% who identified as having a race/ethnicity other than White. Via latent profile analysis (LPA), a three‐group model was identified as best fit to the data. Profile identification was dependent on a participant's combination of cognitive, expressive, and social communication characteristics, rather than a single domain. One group (n = 206) had global delays, while the other two groups (n = 95 and n = 43) had variable strengths in cognition and communication. Findings suggest that low‐frequency/minimally verbal communicators with autism have heterogeneous characteristics that can be systematically organized.