[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-273545-105":59,"doc-detail-273545-en":131},{"code":4,"msg":5,"data":6},0,"success",[7,13,18,23,28,33,38,43,48,51,55],{"id":8,"doc_module":4,"doc_module_name":9,"category_name":10,"show_sort_weight":11,"slug":12},1,"Document","Story & Novel",90,"story-novel",{"id":14,"doc_module":4,"doc_module_name":9,"category_name":15,"show_sort_weight":16,"slug":17},2,"Literature",80,"literature",{"id":19,"doc_module":4,"doc_module_name":9,"category_name":20,"show_sort_weight":21,"slug":22},4,"Exam",70,"exam",{"id":24,"doc_module":4,"doc_module_name":9,"category_name":25,"show_sort_weight":26,"slug":27},5,"Comic",60,"comic",{"id":29,"doc_module":4,"doc_module_name":9,"category_name":30,"show_sort_weight":31,"slug":32},6,"Technology",50,"technology",{"id":34,"doc_module":4,"doc_module_name":9,"category_name":35,"show_sort_weight":36,"slug":37},7,"Healthcare",40,"healthcare",{"id":39,"doc_module":4,"doc_module_name":9,"category_name":40,"show_sort_weight":41,"slug":42},8,"Research & Report",30,"research-report",{"id":44,"doc_module":4,"doc_module_name":9,"category_name":45,"show_sort_weight":46,"slug":47},9,"Religion & Spirituality",20,"religion-spirituality",{"id":46,"doc_module":4,"doc_module_name":9,"category_name":49,"show_sort_weight":46,"slug":50},"World Cup","world-cup",{"id":52,"doc_module":4,"doc_module_name":9,"category_name":53,"show_sort_weight":52,"slug":54},10,"Lifestyle","lifestyle",{"id":56,"doc_module":4,"doc_module_name":9,"category_name":57,"show_sort_weight":24,"slug":58},19,"General","general",{"code":4,"msg":60,"data":61},"ok",{"site_id":62,"language":63,"slug":64,"title":65,"keywords":66,"description":67,"schema_data":68,"social_meta":124,"head_meta":126,"extra_data":128,"updated_unix":130},105,"en","bear-bench-a-bilingual-enterprise-and-academic-reasoning-benchmark-for-multimodal-models-research-and-evaluation","BEAR-Bench - A Bilingual Enterprise and Academic Reasoning Benchmark for Multimodal Models - research and evaluation","","Multimodal Large Language Models show rapid progress in visual comprehension, yet their reasoning over text-dense professional documents remains insufficiently benchmarked. Existing resources often stress OCR, depend on external domain knowledge, or focus mainly on English and Chinese, leaving Russian notably underrepresented. BEAR-Bench introduces a self-contained benchmark with 1000 human-annotated questions spanning business and scientific documents in English and Russian. Sixteen MLLMs are evaluated, and outputs are used to compare hallucination detection methods and their reliability.",{"@graph":69,"@context":123},[70,84,106],{"@type":71,"itemListElement":72},"BreadcrumbList",[73,77,79,82],{"item":74,"name":75,"@type":76,"position":8},"https://docshare.wps.com","Home","ListItem",{"item":78,"name":9,"@type":76,"position":14},"https://docshare.wps.com/document/",{"item":80,"name":40,"@type":76,"position":81},"https://docshare.wps.com/document/research-report/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/bear-bench-a-bilingual-enterprise-and-academic-reasoning-benchmark-for-multimodal-models-research-and-evaluation/273545/",{"url":83,"name":65,"@type":85,"image":86,"author":91,"headline":65,"publisher":94,"fileFormat":97,"inLanguage":63,"description":67,"dateModified":98,"datePublished":99,"encodingFormat":97,"isAccessibleForFree":100,"interactionStatistic":101},"DigitalDocument",{"url":87,"@type":88,"width":89,"height":90},"https://docshare.wps.com/thumbnails/bear-bench-a-bilingual-enterprise-and-academic-reasoning-benchmark-for-multimodal-models-research-and-evaluation/273545.png","ImageObject",300,407,{"name":92,"@type":93},"Maeve","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-10-29","2026-09-15",true,{"@type":102,"interactionType":103,"userInteractionCount":105},"InteractionCounter",{"@type":104},"ViewAction",12,{"@type":107,"mainEntity":108},"FAQPage",[109,115,119],{"name":110,"@type":111,"acceptedAnswer":112},"What problem does BEAR-Bench target in multimodal model evaluation?","Question",{"text":113,"@type":114},"It evaluates reasoning on text-rich, professional documents, where existing benchmarks often focus on OCR, require external knowledge, or do not adequately cover Russian.","Answer",{"name":116,"@type":111,"acceptedAnswer":117},"How is BEAR-Bench designed so questions do not rely on external expertise?",{"text":118,"@type":114},"Questions follow design principles requiring multiple logical steps and ensuring they are fully answerable from the document alone.",{"name":120,"@type":111,"acceptedAnswer":121},"How are hallucination detection methods assessed using BEAR-Bench?",{"text":122,"@type":114},"Model outputs are used to compare token-level uncertainty, representation-based detectors, supervised hidden-state probes, and LLM-as-a-judge approaches across deployment settings.","https://schema.org",{"og:url":83,"og:type":125,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":127,"canonical":83},"index,follow",{"doc_id":129,"site_id":62},273545,1789471995,{"code":4,"msg":5,"data":132},{"doc_id":129,"user_id":133,"nickname":92,"user_avatar":134,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":135,"file_id":136,"file_url":137,"file_type":138,"file_size":139,"view_count":105,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":140,"language":141,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":142,"faqs":143,"seo_title":144,"seo_description":67,"update_tm":130,"read_time":145,"preview_page_count":4},5909877438554,"https://ap-avatar.wpscdn.com/avatar/5600025385ad2bf12a7?_k=1778553567797529272","BEAR-Bench: A Bilingual Enterprise and Academic Reasoning Benchmark for Multimodal Models  \nLiubov Chubarova1 , Alexandra Kuleshova2 , Daniil Volkov∗2, Kirill Sultanov∗1, Alexey Zaytsev2 ,  \n1 Yandex 2 Applied AI Institute  \nCorrespondence: [bazarovaai.239@gmail.com](bazarovaai.239@gmail.com)  \narXiv : 2608 . 17895v 1 [cs .CL] 18 Aug 2026  \nAbstract  \nWhile Multimodal Large Language Models (MLLMs) have made significant strides in visual comprehension, their ability to reason about text-dense, professional documents remains incompletely evaluated. Existing benchmarks emphasize information extraction, require external domain knowledge, or cover professional documents only as one of many settings. They are also largely English- or Chinese-centric, leaving other languages and Russian, in particular, substantially underrepresented. To address these limitations, we introduce BEAR-Bench (Bilingual Enterprise and Academic Reasoning), a self-contained, complex English-and-Russian benchmark comprising 1000 human-annotated questions based on text-rich business and scientific documents. We evaluate 16 proprietary and openweight MLLMs, including Gemini 3.1 Pro and Qwen3.5-397B, on BEAR-Bench and observe clear headroom even for the strongest systems. Finally, we use the resulting model outputs to compare existing hallucination detection methods, evaluating not only how often models fail on BEAR-Bench but also how reliably those failures can be identified.  \n1 Introduction  \nMultimodal Large Language Models (MLLMs) (OpenAI et al., 2024 ; Team et al., 2025 ; Anthropic, 2024) have revolutionized the machine comprehension of images. Beyond merely solving optical character recognition (OCR) tasks, these models demonstrate the potential to perform complex reasoning over multimodal inputs (Yue et al., 2024 ; Hao et al., 2025) . This capability is particularly vital for text-rich scenarios, which are central to real-world professional domains that require reading and analyzing dense visual documents, such as scientific papers and financial reports.  \n*  \nEqual contribution.  \nExisting benchmarks provide complementary but incomplete coverage of professional document reasoning. Document-oriented datasets such as DocVQA (Mathew et al., 2021) primarily emphasize OCR and information extraction, whereas broad multimodal reasoning benchmarks such as MMMU (Yue et al., 2024) often require specialized factual knowledge. OCR-Reasoning (Huang et al., 2026) covers diverse text-rich images, including some professional documents, but does not examine this setting in depth. A further limitation is linguistic coverage: existing multimodal benchmarks exhibit a strong bias toward English and Chinese, leaving Slavic languages—and Russian in particular—substantially underrepresented. Russianinclusive resources such as MTVQA (Tang et al., 2025), MWS Vision Bench (MWS AI, 2025), and MERA-Multi (Chervyakov et al., 2026) broaden this coverage, but mix documents with other image types, emphasize OCR and document processing, or focus on a narrow document category. To our knowledge, no existing benchmark evaluates self-contained, multi-step reasoning over Russianlanguage professional document pages whose items involve textual and graphical content such as figures, tables, charts, equations, and diagrams.  \nTo address these limitations, we introduce BEAR-Bench, a complex benchmark comprising 1000 human-annotated questions for text-dense enterprise and scientific documents in English and Russian languages. The science section tests the ability to interpret academic figures, transcribe mathematical formulas, and analyze data from plots. The enterprise tasks target data crossreferencing in financial reports and logical reasoning over business charts. The questions follow two design principles. First, they require multiple logical steps rather than direct extraction from a single piece of evidence. Second, they are fully answerable from the document alone, without external expert ","cbCaighFMuXaU6t5","https://ap.wps.com/l/cbCaighFMuXaU6t5","pdf",3843816,17,"English","# Abstract\n# Introduction\n## Multimodal reasoning over professional documents\n## Linguistic coverage gaps and motivation\n## BEAR-Bench design principles and coverage\n## Hallucination detection evaluation\n# Related Work\n## Multimodal benchmarks for professional documents","[{\"question\":\"What problem does BEAR-Bench target in multimodal model evaluation?\",\"answer\":\"It evaluates reasoning on text-rich, professional documents, where existing benchmarks often focus on OCR, require external knowledge, or do not adequately cover Russian.\"},{\"question\":\"How is BEAR-Bench designed so questions do not rely on external expertise?\",\"answer\":\"Questions follow design principles requiring multiple logical steps and ensuring they are fully answerable from the document alone.\"},{\"question\":\"How are hallucination detection methods assessed using BEAR-Bench?\",\"answer\":\"Model outputs are used to compare token-level uncertainty, representation-based detectors, supervised hidden-state probes, and LLM-as-a-judge approaches across deployment settings.\"}]","BEAR-Bench - A Bilingual Enterprise and Academic Reasoning Benchmark for Multimodal Models - research and evaluation | PDF",43]