Our scoring engine powered by DeepSeek V3 delivers expert-level accuracy at machine speed -- with detailed, actionable feedback on every essay.
Powered by DeepSeek V3, our engine scores essays and generates detailed, personalized feedback -- quoting student work and suggesting specific improvements.
QWK 0.77 achieved in benchmark testing on 24,728 expert-graded essays (ASAP 2.0 dataset). Benchmark results reflect testing conditions and are not guaranteed in all production scenarios.
Quadratic Weighted Kappa (QWK) is the standard metric for evaluating automated essay scoring systems. It measures agreement between the automated scores and expert human graders, penalizing larger disagreements more heavily than smaller ones.