Üstün and Aryabumi teams: Aya multilingual instruction and evaluation resource

What changed

Released instruction-tuned multilingual models with multilingual data mixtures and expanded evaluation coverage; compared broad 101-language coverage with a later 23-language depth-focused allocation using the same instruction resource.

Scientific scope

Released instruction-tuned multilingual models with multilingual data mixtures and expanded evaluation coverage; compared broad 101-language coverage with a later 23-language depth-focused allocation using the same instruction resource. One resource/contribution package, not a new invention for each language count or model size. Native people list is the union of two separately documented bylines, not one jointly authored paper. Aya23 reuses the initial instruction mixture and evaluation framework; licensing and language-specific performance limitations remain.

Evidence

The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. One resource/contribution package, not a new invention for each language count or model size. Native people list is the union of two separately documented bylines, not one jointly authored paper. Aya23 reuses the initial instruction mixture and evaluation framework; licensing and language-specific performance limitations remain.

Contributors

• Source-listed authors and technical contributors · SCIENCE: Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.

• Aya Model: An Instruction Finetuned Open-Access Multilingual Language Model
• Üstün and Aryabumi teams: Aya multilingual instruction and evaluation resource — official research provenance
• Aya 23: Open Weight Releases to Further Multilingual Progress

Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.

Full scientific record · Browse contribution grades · Request a correction