The third generation of the Hungarian National Corpus (MNSZ3) aims to provide a large-scale, curated, and well-described corpus resource needed for the sustainable digital presence of Hungarian. Building on the domain structure and proportions of MNSZ2 (v2.0.5; 1.04 billion running words), the project targets a substantial increase in scale while also strengthening the coverage and metadata description of Hungarian language use outside Hungary. MNSZ3 retains the six traditional domains of the earlier corpus—press, fiction, scientific, official, personal, and transcribed spoken language—and is planned to reach approximately 10 billion tokens. This paper presents the motivation and design principles of the project, outlines the practical decisions and procedures used in data collection and cleaning, and discusses the annotation strategy developed for large-scale processing. In planning the linguistic analysis, we build on the complementary strengths of HuSpaCy and e-magyar: HuSpaCy provides the unified and efficient UD-oriented processing backbone, while e-magyar (emMorph) is preserved as an explicit additional layer for morphology and lemmatisation.

In this study, we introduce the Hungarian Generative Model Evaluation (HuGME) benchmark, a new framework designed to assess the linguistic proficiency of large language models (LLMs) in Hungarian. HuGME evaluates models across a diverse set of linguistic and reasoning skills, including bias, toxicity, faithfulness, relevance, summarization, prompt alignment, readability, spelling, grammaticality, and domain-specific knowledge through tasks like TruthfulQA and MMLU. We applied HuGME to a range of Hungarian LLMs, including those developed in-house as well as several publicly available models that claim Hungarian language proficiency. This paper presents the comparative results of these evaluations, shedding light on the capabilities of current LLMs in processing the Hungarian language. Through our analysis, we aim to both showcase the current state of Hungarian linguistic processing in LLMs and provide a foundational resource for future advancements in the field.

We introduce OpenHuEval, the first benchmark for LLMs focusing on the Hungarian language and specifics. OpenHuEval is constructed from a vast collection of Hungarian-specific materials sourced from multiple origins. In the construction, we incorporated the latest design principles for evaluating LLMs, such as using real user queries from the internet, emphasizing the assessment of LLMs’ generative capabilities, and employing LLM-as-judge to enhance the multidimensionality and accuracy of evaluations. Ultimately, OpenHuEval encompasses eight Hungarian-specific dimensions, featuring five tasks and 3953 questions. Consequently, OpenHuEval provides the comprehensive, in-depth, and scientifically accurate assessment of LLM performance in the context of the Hungarian language and its specifics. We evaluated current mainstream LLMs, including both traditional LLMs and recently developed Large Reasoning Models. The results demonstrate the significant necessity for evaluation and model optimization tailored to the Hungarian language and specifics. We also established the framework for analyzing the thinking processes of LRMs with OpenHuEval, revealing intrinsic patterns and mechanisms of these models in non-English languages, with Hungarian serving as a representative example. We will release OpenHuEval at https://github.com/opendatalab/OpenHuEval .