SmartILR

AI-powered ILR passage rating for language teachers

AI ILR Evaluation

SmartILR applies AI to the ILR evaluation process — not to replace the teacher's judgment, but to do the analytical work that takes the most time: reading a passage carefully, scoring it across multiple dimensions, and producing a written justification that holds up to scrutiny.

Teachers who use SmartILR spend less time on the mechanics of level placement and more time on instruction. The AI handles the first pass; the teacher reviews, adjusts, and owns the decision.

How SmartILR's AI evaluation works

Input your passage or audio Paste a reading passage, upload a document, or submit an audio or video file for listening evaluation
AI scores across ILR dimensions SmartILR evaluates discourse organization, vocabulary difficulty, inferential demand, syntactic complexity, and delivery factors for audio
Level placement with justification An ILR level is assigned with a three-part written justification: why this level, why not higher, and why not lower
Export or review the full report Copy the complete analysis for documentation, share it with a colleague, or use it to inform a curriculum decision

What the AI does not do

SmartILR's AI evaluation is not a certified ILR rating. It does not replace expert raters in formal test development contexts, and it does not produce scores that carry the same weight as panel-reviewed ILR ratings. It is designed to support teacher workflows — not to substitute for official proficiency certification processes.

AI models and reliability

SmartILR uses GPT-4o for reading and text-based listening analysis, and a dedicated audio model for direct audio scoring. All models run at temperature 0 for maximum consistency. The same passage submitted twice will receive the same level placement. Confidence labels and numeric scores indicate where the AI considers the placement strong versus borderline.

Teachers working across Arabic, English, French, Spanish, and other languages can use SmartILR with the same workflow — language detection is automatic, or can be set manually for ambiguous samples.

Start an AI ILR Evaluation