We are pleased to announce SemEval-2026 Task 7: Everyday Knowledge Across Diverse Languages and Cultures.
This task extends BLEnD (Myung et al., 2024), a hand-crafted benchmark built by native speakers to evaluate cultural knowledge in large language models (LLMs) and NLP systems,to more than 30 language-culture pairs. The goal is to evaluate models and provide insights for improving cross-cultural performance.
BLEnD’s manually constructed data is used only for validation and testing, ensuring that models are assessed on their ability to generalise to unseen cultural and linguistic contexts.
We invite researchers, practitioners, and students to participate. Note that our task is inclusive and junior-friendly. We will organise a live Q&A session and a system-description writing tutorial for all participants.
**Tracks**
- Track 1: Short Answer Questions (SAQ) – Given a question in a target language, the model must generate an answer in that same language.
- Track 2: Multiple Choice Questions (MCQ) – Each question is in English, with four answer options reflecting different cultural perspectives (countries or regions).
**Contact **
- Competition website: https://www.codabench.org/competitions/10281/
- Discord channel (updates & discussion): https://discord.com/invite/8x6XP97kmw
- GitHub: https://github.com/BLEnD-SemEval2026/SemEval-2026-Task-7/
- Email organisers: semeval-2026-blend-organisers [at] googlegroups.com
- Participants’ Google Group: semeval-2026-task7-blend-participants [at] googlegroups.com
**Important dates (subject to change)**
- Sample data ready: 8 August 2025
- Evaluation start: 10 January 2026
- Evaluation end: 31 January 2026
- Paper submission due: ~February 2026
- SemEval workshop: Summer 2026 (co-located with a major NLP conference)
Best regards,
The SemEval-2026 Task 7 Organising Team.