The grader evaluates students’ practice writing against the state’s published rubrics and returns a score and feedback in about three seconds.

CoGrader, the AI essay grader trusted by more than 100,000 teachers, today released a study that shows its Texas STAAR AI Grader scored essays within one point of the official Texas Education Agency (TEA) score on 98.3% of 1,764 STAAR constructed responses.

This press release features multimedia. View the full release here: https://www.businesswire.com/news/home/20261109150003/en/

Data from a new study shows that CoGrader’s Texas STAAR AI Grader exactly matches a human grader’s score as much as 90.5% of the time.

“The within-one-point threshold is TEA’s standard for two human raters, and a fair test for an automated scorer is whether it agrees with a trained rater about as often as a second trained rater would,” said Gil Flores, founder and CEO of CoGrader. “This data shows that the Texas STAAR AI Grader earned an A on that test.”

CoGrader’s grading engine was also able to exactly match the TEA-assigned rating on 80.4% of the responses. For reference, the CoGrader study found that two trained TEA raters had exactly matching scores 63% to 75% of the time.

The student responses were drawn from released TEA scoring guides and rater-training materials covering grades 3 to 8, English I, English II, Biology, and U.S. History. The redesigned STAAR includes both Short Constructed Responses (SCRs) and Extended Constructed Responses (ECRs). When scoring SCRs, the Texas STAAR AI Grader was within one point of the TEA rating 99.7% of the time. For ECRs, the grader’s score was within one point of the TEA rating 96.0% of the time.

The Texas STAAR AI Grader uses the publicly released TEA STAAR scoring guides as in-context calibration material. The grader evaluates student practice writing against the state’s published rubrics and returns a score with rubric-aligned feedback in approximately three seconds. The teacher then reviews each score, adjusts the feedback where needed, and decides exactly what each student sees.

Although the spring operational exam is assessed by machines, students’ classroom and benchmark writing throughout the school year is not. This leaves teachers to manually score every practice essay, benchmark, and short answer. Without CoGrader, a teacher might take 5 minutes to grade one essay. For 130 students, that adds up to 10 hours. With CoGrader, teachers can grade an essay in 2 or 3 minutes, saving them as much as 6 hours per assignment.

“These results demonstrate that the CoGrader AI Grader not only saves teachers time, but matches the reliability of the state’s own human rater pool,” concluded Flores.

About CoGrader

CoGrader is the AI essay grader trusted by more than 100,000 teachers and impacting more than 1,000,000 students in US schools. Research on CoGrader was funded by an IES grant from the U.S. Department of Education (Award R305J250071). Built on decades of research on writing feedback and revision, the grader evaluates students’ practice writing against state, AP, or custom rubrics and returns a score, along with consistently objective feedback, in approximately three seconds. Teachers review each score, personalize the feedback, and decide exactly what each student sees. Learn more at CoGrader.com.

View source version on businesswire.com: https://www.businesswire.com/news/home/20261109150003/en/

“A fair test for an automated scorer is whether it agrees with a trained rater about as often as a second trained rater would,” said Gil Flores, founder and CEO of CoGrader. “This data shows that the Texas STAAR AI Grader earned an A on that test.”