Open CEFR is an open-source, zero-dependency CEFR (A1–C2) proficiency assessment platform featuring dynamic item difficulty routing, Wilson confidence intervals, and a 93-question curated open dataset.
A1–C2
Complete CEFR proficiency spectrum
96.3%
Within ±1 level accuracy across 1,000 synthetic test-takers
0 Dependencies
Pure TypeScript ESM package for Node, Browser & Edge
Key Pain Point: Static placement tests require 40+ questions to achieve reliability, inducing significant test-taker fatigue.
Engineered Open CEFR with psychometric item routing, pattern-accelerated ±2 level fast-tracking, 95% Wilson score binomial confidence bounds, recency-weighted scoring, and a dual-licensed dataset (MIT + CC BY 4.0).
The zero-dependency TypeScript architecture ensures out-of-the-box compatibility with Node.js, Web Browsers, Cloudflare Workers, and React Native.
Curated dataset includes a 36-question core adaptive pool and 57 specialized subject-track questions spanning 19 diagnostic modules.
A comprehensive CEFR curriculum graph maps A1 to C2 topic prerequisites with communicative starter prompts.
Open CEFR is a standardized, zero-dependency psychometric English assessment engine, 93-question curated CEFR dataset (A1–C2), and interactive web runner. Developed by Epheos to empower educators, EdTech startups, and schools, it delivers accurate level placement in under 12 questions with full sub-skill diagnostics and zero proprietary vendor lock-in.
Our research revealed critical insights that informed our strategy and implementation.
The zero-dependency TypeScript architecture ensures out-of-the-box compatibility with Node.js, Web Browsers, Cloudflare Workers, and React Native.
Curated dataset includes a 36-question core adaptive pool and 57 specialized subject-track questions spanning 19 diagnostic modules.
A comprehensive CEFR curriculum graph maps A1 to C2 topic prerequisites with communicative starter prompts.
Dual-licensing under MIT (code) and Creative Commons CC BY 4.0 (educational datasets) fosters widespread community contribution.
Monte Carlo population benchmarks with 1,000 synthetic test-takers confirmed 96.3% placement accuracy within ±1 CEFR level.
Open CEFR establishes an open standard for language proficiency assessment, democratizing adaptive placement testing for educators and developers worldwide.
A clear roadmap for implementing these strategies and next steps.
Install open-cefr-exam via npm or clone the repository from GitHub.
Initialize CEFRExamSession with the curated default question pool or custom topic tracks.
Integrate the session dispatcher into web assessment interfaces or mobile onboarding flows.
Utilize Wilson confidence score metrics to trigger early exam termination.
Generate personalized diagnostic reports and CEFR level certificates for candidates.
Get expert guidance implementing these strategies
Let's work together to achieve remarkable results for your organization.
