benchmark_id stringlengths 12 56 | slug stringlengths 3 56 | name stringlengths 2 44 | source stringclasses 4
values | source_url stringlengths 19 110 | description stringlengths 0 995 | categories listlengths 0 21 | languages listlengths 0 13 | modality stringclasses 5
values | publisher stringlengths 3 80 ⌀ | released_at timestamp[s]date 2010-02-19 00:00:00 2026-08-25 00:00:00 ⌀ | openness stringclasses 3
values | paper_url stringlengths 31 62 ⌀ | repo_url stringlengths 28 89 ⌀ | dataset_url stringlengths 40 81 ⌀ | document_count int64 1 27 ⌀ | model_count int64 1 586 ⌀ | score_count int64 0 586 | highest_score float64 0.01 2.1M ⌀ | score_unit stringclasses 3
values |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
llm-stats:vibe-eval | llm-stats-vibe-eval | Vibe-Eval | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/vibe-eval?top_n=500 | VIBE-Eval is a hard evaluation suite for measuring progress of multimodal language models, consisting of 269 visual understanding prompts with gold-standard responses authored by experts. The benchmark has dual objectives: vibe checking multimodal chat models for day-to-day tasks and rigorously testing frontier models,... | [
"multimodal",
"general",
"vision"
] | [] | multimodal | null | null | unknown | null | null | null | 1 | 8 | 8 | 0.672 | null |
llm-stats:vibe-pro | llm-stats-vibe-pro | VIBE-Pro | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/vibe-pro?top_n=500 | VIBE-Pro is an advanced version of the VIBE (Visual & Interactive Benchmark for Execution) benchmark that evaluates LLMs on professional-grade full-stack application development tasks. It measures model performance across complex real-world development scenarios including web, mobile, and backend applications with high... | [
"agents",
"code"
] | [] | text | null | null | unknown | null | null | null | 1 | 2 | 2 | 0.556 | null |
llm-stats:vibe-v2 | llm-stats-vibe-v2 | VIBE-V2 | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/vibe-v2?top_n=500 | VIBE-V2 is an internal benchmark covering pure front-end and full-stack Web, Android, and iOS projects with build-from-scratch tasks. It uses an Agent-as-a-Verifier paradigm to automatically verify program interaction logic and visual output, scoring models through a unified pipeline that includes a requirement set, co... | [
"agents",
"code"
] | [] | text | null | null | unknown | null | null | null | 1 | 1 | 1 | 0.5012 | null |
model-reports:vibench | vibench | ViBench | model_reports | https://vibench.ai/ | End-to-end vibe-coding benchmark by authors at Replit and Georgian AI Lab, scoring web applications from the user's perspective. Tasks are derived from anonymized Replit production traces, so the benchmark is first-party to one of the agents it measures. It reaches this registry through a customer testimonial in Anthro... | [
"coding_agent"
] | [] | null | null | 2026-05-26T00:00:00 | unknown | null | null | null | 1 | null | 0 | null | null |
llm-stats:video-mme | llm-stats-video-mme | Video-MME | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/video-mme?top_n=500 | Video-MME is the first-ever comprehensive evaluation benchmark of Multi-modal Large Language Models (MLLMs) in video analysis. It features 900 videos totaling 254 hours with 2,700 human-annotated question-answer pairs across 6 primary visual domains (Knowledge, Film & Television, Sports Competition, Life Record, Multil... | [
"multimodal",
"reasoning",
"vision"
] | [] | multimodal | null | null | unknown | null | null | null | 1 | 17 | 17 | 0.892 | null |
model-reports:video_mme | video_mme | Video-MME | model_reports | https://video-mme.github.io/ | Frame sampling rate dominates long-video results. | [
"multimodal"
] | [] | null | null | 2024-05-31T00:00:00 | unknown | null | null | null | 4 | 2 | 3 | 90 | percent |
opencompass:1358 | opencompass-1358-video-mme | Video-MME | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/Video-MME | Video-MME is an evaluation benchmark of multi-modal LLMs in video analysis, including 900 videos in various duration with a total of 254 hours which spans 6 primary visual domains with 30 subfields. Video-MME用于评估多模态大模型的视频分析能力，包含900个不同长度的视频，来自6个主要视觉领域和30个子领域，总时长达254小时。 | [
"多模态",
"Multimodal",
"理解",
"Understanding",
"多模态模型",
"VLM",
"视频理解",
"Video Understanding",
"不支持",
"Unsupported"
] | [] | multimodal | University of Science and Technology of China | 2024-03-31T00:00:00 | unknown | null | https://github.com/BradyFU/Video-MME | null | 1 | null | 0 | null | null |
llm-stats:video-mme-(long,-no-subtitles) | llm-stats-video-mme-long-no-subtitles | Video-MME (long, no subtitles) | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/video-mme-%28long%2C-no-subtitles%29?top_n=500 | Video-MME is the first-ever comprehensive evaluation benchmark for Multi-modal Large Language Models (MLLMs) in video analysis. This variant focuses on long-term videos (30min-60min) without subtitle inputs, testing robust contextual dynamics across 6 primary visual domains with 30 subfields including knowledge, film &... | [
"multimodal",
"video",
"vision"
] | [] | multimodal | null | null | unknown | null | null | null | 1 | 1 | 1 | 0.72 | null |
opencompass:1277 | opencompass-1277-videogui | VideoGUI | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/VideoGUI | VideoGUI is designed to evaluate GUI assistants on visual-centric GUI tasks. Sourced from high-quality web instructional videos, it focuses on tasks involving professional and novel software and complex activities (e.g., video editing). VideoGUI旨在评估以视觉为中心的GUI任务上的GUI助手，来自高质量的网络教学视频，侧重于涉及专业和新颖软件和复杂活动（例如视频编辑）的任务。 | [
"创作",
"Creation",
"NeurIPS 2024",
"智能体",
"Agent",
"任务执行",
"Task Execution",
"视频理解",
"Video Understanding",
"不支持",
"Unsupported"
] | [] | null | National University of Singapore | 2024-06-14T00:00:00 | unknown | https://arxiv.org/abs/2406.10227 | https://github.com/showlab/videogui | null | 1 | null | 0 | null | null |
llm-stats:videoholmes | llm-stats-videoholmes | VideoHolmes | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/videoholmes?top_n=500 | VideoHolmes evaluates video understanding and reasoning capabilities in multimodal models. | [
"multimodal",
"reasoning",
"video"
] | [] | multimodal | null | null | unknown | null | null | null | 1 | 3 | 3 | 0.682 | null |
opencompass:1926 | opencompass-1926-videomathqa | VideoMathQA | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/VideoMathQA | VideoMathQA is a benchmark designed to evaluate mathematical reasoning in real-world educational videos. It requires models to interpret and integrate information from three modalities, visuals, audio, and text, across time. VideoMathQA是一个旨在评估实际教育视频中数学推理能力的基准。它要求模型解释和整合来自三种模态(视觉、音频和文本)随时间变化的信息。该基准解决了"多模态针堆"问题,即关键信息稀疏且分... | [
"多模态",
"Multimodal",
"推理",
"Reasoning",
"多模态模型",
"VLM",
"逻辑推理",
"视频理解",
"Video Understanding",
"不支持",
"Unsupported"
] | [] | multimodal | MBZUAI,University of California Merced,Google Research,etc | 2025-06-05T00:00:00 | unknown | https://arxiv.org/abs/2506.05349 | https://github.com/mbzuai-oryx/VideoMathQA | null | 1 | null | 0 | null | null |
llm-stats:videomme-w-sub. | llm-stats-videomme-w-sub | VideoMME w sub. | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/videomme-w-sub.?top_n=500 | The first-ever comprehensive evaluation benchmark of Multi-modal LLMs in Video analysis. Features 900 videos (254 hours) with 2,700 question-answer pairs covering 6 primary visual domains and 30 subfields. Evaluates temporal understanding across short (11 seconds) to long (1 hour) videos with multi-modal inputs includi... | [
"multimodal",
"video",
"vision"
] | [] | multimodal | null | null | unknown | null | null | null | 1 | 10 | 10 | 0.904 | null |
llm-stats:videomme-w-o-sub. | llm-stats-videomme-w-o-sub | VideoMME w/o sub. | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/videomme-w-o-sub.?top_n=500 | Video-MME is a comprehensive evaluation benchmark for multi-modal large language models in video analysis. It features 900 videos across 6 primary visual domains with 30 subfields, ranging from 11 seconds to 1 hour in duration, with 2,700 question-answer pairs. The benchmark evaluates MLLMs' capabilities in processing ... | [
"multimodal",
"video",
"vision"
] | [] | multimodal | null | null | unknown | null | null | null | 1 | 10 | 10 | 0.839 | null |
llm-stats:videommmu | llm-stats-videommmu | VideoMMMU | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/videommmu?top_n=500 | Video-MMMU evaluates Large Multimodal Models' ability to acquire knowledge from expert-level professional videos across six disciplines through three cognitive stages: perception, comprehension, and adaptation. Contains 300 videos and 900 human-annotated questions spanning Art, Business, Science, Medicine, Humanities, ... | [
"multimodal",
"reasoning",
"healthcare",
"vision"
] | [] | multimodal | null | null | unknown | null | https://github.com/EvolvingLMMs-Lab/VideoMMMU | null | 1 | 26 | 26 | 0.876 | null |
opencompass:1900 | opencompass-1900-videoreasonbench | VideoReasonBench | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/VideoReasonBench | VideoReasonBench is designed to evaluate vision-centric complex video reasoning. VideoReasonBench是一个用于评测视觉为中心、复杂视频推理的基准。 | [
"强推理",
"Strong Reasoning",
"多模态",
"Multimodal",
"video reasoning",
"MLLMs",
"多模态模型",
"VLM",
"逻辑推理",
"Reasoning",
"视频理解",
"Video Understanding",
"不支持",
"Unsupported"
] | [] | multimodal | Peking University | 2025-06-05T00:00:00 | restricted | https://arxiv.org/pdf/2505.23359 | https://github.com/llyx97/video_reason_bench | https://huggingface.co/datasets/lyx97/reasoning_videos | 1 | null | 0 | null | null |
llm-stats:videosimpleqa | llm-stats-videosimpleqa | VideoSimpleQA | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/videosimpleqa?top_n=500 | VideoSimpleQA evaluates factual knowledge grounded in video content, measuring how accurately models answer short, fact-seeking questions about videos. | [
"multimodal",
"knowledge",
"video",
"vision"
] | [] | multimodal | null | null | unknown | null | null | null | 1 | 2 | 2 | 0.764 | null |
opencompass:1717 | opencompass-1717-vilbench | ViLBench | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/ViLBench | ViLBench is a benchmark designed to evaluate vision-language models. It features 600 examples from 5 datasets, selected based on the criterion that process reward models offer greater improvements over output reward models in guiding generations. ViLBench 是一项旨在评估视觉-语言模型的数据集，其强调对模型进行细粒度的逐步推理能力测试。该基准共包含600个经过严格筛选的样本，来源于五... | [
"多模态",
"Multimodal",
"推理",
"Reasoning",
"vision-language",
"math",
"多模态模型",
"VLM",
"逻辑推理",
"不支持",
"Unsupported"
] | [] | multimodal | UC Santa Cruz | 2025-03-26T00:00:00 | unknown | https://arxiv.org/abs/2503.20271 | null | https://huggingface.co/datasets/UCSC-VLAA/ViLBench | 1 | null | 0 | null | null |
llm-stats:vct | llm-stats-vct | Virology Capabilities Test | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/vct?top_n=500 | Virology Capabilities Test (VCT) is an expert-level multiple-choice benchmark measuring the capability to troubleshoot complex virology laboratory protocols. It evaluates dual-use biological knowledge relevant to bioweapons development. | [
"safety",
"healthcare"
] | [] | text | null | null | unknown | null | null | null | 1 | 1 | 1 | 0.61 | null |
opencompass:2063 | opencompass-2063-visco | VISCO | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/VISCO | VISCO aims to evaluate the critique and correction capabilities of VLMs, which are two essential building blocks towards VLM self-improvement. VISCO requires VLMs to critique the correctness of each step in CoT, provide natural language explanation, and corrects the CoT based on the critique. VISCO 旨在评估 VLM 的评判（critiqu... | [
"多模态",
"Multimodal",
"推理",
"Reasoning",
"知识",
"Knowledge",
"self-critique",
"VLM",
"VLM reasoning",
"多模态模型",
"跨模态推理",
"Cross-modal Reasoning",
"不支持",
"Unsupported"
] | [] | multimodal | UCLA | 2024-12-03T00:00:00 | unknown | https://arxiv.org/abs/2412.02172 | https://github.com/PlusLabNLP/VISCO | https://huggingface.co/datasets/uclanlp/VISCO | 1 | null | 0 | null | null |
llm-stats:visfactor | llm-stats-visfactor | VisFactor | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/visfactor?top_n=500 | VisFactor is a benchmark evaluating fine-grained visual factor perception and reasoning over images. | [
"multimodal",
"reasoning",
"vision"
] | [] | multimodal | null | null | unknown | null | null | null | 1 | 3 | 3 | 0.514 | null |
llm-stats:vision2web | llm-stats-vision2web | Vision2Web | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/vision2web?top_n=500 | Vision2Web evaluates multimodal models on converting visual designs and screenshots into functional web pages, measuring end-to-end design-to-code capability. | [
"multimodal",
"code",
"vision"
] | [] | multimodal | null | null | unknown | null | null | null | 1 | 3 | 3 | 0.69 | null |
opencompass:1993 | opencompass-1993-vistorybench | ViStoryBench | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/ViStoryBench | ViStoryBench is a benchmark designed to rigorously evaluate the capabilities of multimodal generative models (e.g., diffusion models, LLM-based agents) in synthesizing visually coherent image sequences from textual narratives and reference images. ViStoryBench 是一个面向故事可视化任务的综合性评测基准，旨在评估多模态生成模型（如扩散模型视频生成模型等）根据给定叙事文本和参考图像... | [
"多模态",
"Multimodal",
"创作",
"Creation",
"多模态模型",
"VLM",
"视觉生成",
"Visual Generation",
"不支持",
"Unsupported"
] | [
"English",
"Chinese"
] | multimodal | Shanghai Tech University , StepFun , AIGC Research , AGI Lab, Westlake Universit | 2025-06-25T00:00:00 | open | https://arxiv.org/abs/2505.24862 | https://github.com/vistorybench/vistorybench | https://huggingface.co/datasets/ViStoryBench/ViStoryBench | 1 | null | 0 | null | null |
opencompass:1748 | opencompass-1748-visualpuzzles | VisualPuzzles | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/VisualPuzzles | VisualPuzzles is a benchmark that targets visual reasoning while deliberately minimizing reliance on specialized knowledge. VisualPuzzles consists of 1168 diverse questions spanning five categories: algorithmic, analogical, deductive, inductive, and spatial reasoning. LLM 能考公务员吗？我们做了个测试…
近年来，大模型（LLM）的能力突飞猛进，似乎“越来越聪明”了... | [
"强推理",
"Strong Reasoning",
"多模态",
"Multimodal",
"推理",
"Reasoning",
"多模态模型",
"VLM",
"逻辑推理",
"不支持",
"Unsupported"
] | [] | multimodal | Carnegie Mellon University | 2025-04-16T00:00:00 | unknown | https://arxiv.org/pdf/2504.10342 | https://github.com/neulab/VisualPuzzles | https://huggingface.co/datasets/neulab/VisualPuzzles | 1 | null | 0 | null | null |
opencompass:1634 | opencompass-1634-visualsimpleqa | VisualSimpleQA | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/VisualSimpleQA | VisualSimpleQA is a multimodal fact-seeking benchmark with two key features. VisualSimpleQA 是一个多模态事实寻求基准，具有两个关键特性。首先，它使视觉和语言模态中 LVLMs 的评估更加简化和解耦。其次，它纳入了明确的难度标准，以指导人工标注并促进提取具有挑战性的子集，即 VisualSimpleQA-hard。 | [
"多模态",
"Multimodal",
"多模态模型",
"VLM",
"跨模态推理",
"Cross-modal Reasoning",
"事实可靠性",
"Factual Reliability",
"不支持",
"Unsupported"
] | [] | multimodal | Zhongguancun Laboratory, RUC, Tencent, etc. | 2025-03-09T00:00:00 | restricted | https://arxiv.org/pdf/2503.06492 | null | https://huggingface.co/datasets/WYLing/VisualSimpleQA | 1 | null | 0 | null | null |
llm-stats:visualwebbench | llm-stats-visualwebbench | VisualWebBench | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/visualwebbench?top_n=500 | A multimodal benchmark designed to assess the capabilities of multimodal large language models (MLLMs) across web page understanding and grounding tasks. Comprises 7 tasks (captioning, webpage QA, heading OCR, element OCR, element grounding, action prediction, and action grounding) with 1.5K human-curated instances fro... | [
"multimodal",
"frontend_development",
"vision"
] | [] | multimodal | null | null | unknown | null | null | null | 1 | 2 | 2 | 0.797 | null |
llm-stats:visulogic | llm-stats-visulogic | VisuLogic | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/visulogic?top_n=500 | VisuLogic evaluates logical reasoning capabilities in visual contexts. | [
"multimodal",
"reasoning",
"vision"
] | [] | multimodal | null | null | unknown | null | null | null | 1 | 3 | 3 | 0.543 | null |
llm-stats:vita-bench | llm-stats-vita-bench | VITA-Bench | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/vita-bench?top_n=500 | VITA-Bench evaluates AI agents on real-world virtual task automation, measuring their ability to complete complex multi-step tasks in simulated environments. | [
"reasoning",
"agents"
] | [] | text | null | null | unknown | null | null | null | 1 | 10 | 10 | 0.497 | null |
opencompass:2388 | opencompass-2388-vknowu | VKnowU | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/VKnowU | 尽管多模态大型语言模型（MLLMs）在识别物体方面已相当熟练，但它们往往缺乏对世界潜在物理和社会原则的直觉的、类似人类的理解。这种高级的、基于视觉的语义，我们称之为视觉知识，其构成了感知与推理之间的桥梁。为了系统地评估这种能力，我们提出了 VKnowU，一个包含 1,249 个视频、1,680 个问题的综合基准，涵盖了 8 种核心类型的视觉知识，既包括以世界为中心的（例如直觉物理），也包括以人类为中心的任务（例如主观意图）。 尽管多模态大型语言模型（MLLMs）在识别物体方面已相当熟练，但它们往往缺乏对世界潜在物理和社会原则的直觉的、类似人类的理解。这种高级的、基于视觉的语义，我们称之为视觉知识，其构成了感知与推理之间的桥梁。为了系... | [
"多模态",
"Multimodal",
"推理",
"Reasoning",
"视频理解",
"Video-Understanding",
"Visual Knowledge",
"Physics",
"Psychology",
"物理智能",
"Embodied AI",
"逻辑推理",
"Video Understanding",
"不支持",
"Unsupported"
] | [] | multimodal | Shanghai AI Laboratory | 2025-11-25T00:00:00 | restricted | https://github.com/OpenGVLab/VKnowU | https://huggingface.co/datasets/Eurayka/VKnowU | 1 | null | 0 | null | null | |
llm-stats:vladbench | llm-stats-vladbench | VLADBench | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/vladbench?top_n=500 | VLADBench is a vision-language autonomous-driving benchmark evaluating understanding of dynamic traffic scenes and participants. | [
"multimodal",
"reasoning",
"vision"
] | [] | multimodal | null | null | unknown | null | null | null | 1 | 1 | 1 | 0.772 | null |
opencompass:1541 | opencompass-1541-vlm2-bench | VLM2-Bench | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/VLM2-Bench | VLM²-Bench is the first comprehensive benchmark that evaluates vision-language models' (VLMs) ability to visually link matching cues across multi-image sequences and videos. The benchmark consists of 9 subtasks with over 3,000 test cases. VLM²-Bench 是第一个全面评估视觉语言模型（VLMs）在多图像序列和视频中视觉链接匹配线索能力的基准。该基准包括 9 个子任务，超过 3000 个测试案例... | [
"多模态",
"Multimodal",
"多模态模型",
"VLM",
"跨模态推理",
"Cross-modal Reasoning",
"不支持",
"Unsupported"
] | [] | multimodal | HKUST; CMU; MIT | 2025-02-17T00:00:00 | restricted | https://arxiv.org/abs/2502.12084 | https://github.com/vlm2-bench/VLM2-Bench | https://huggingface.co/datasets/Sterzhang/vlm2-bench | 1 | null | 0 | null | null |
llm-stats:vlmsarebiased | llm-stats-vlmsarebiased | VLMsAreBiased | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/vlmsarebiased?top_n=500 | VLMsAreBiased evaluates whether vision-language models rely on visual evidence or fall back on language priors when answering. | [
"multimodal",
"vision"
] | [] | multimodal | null | null | unknown | null | null | null | 1 | 2 | 2 | 0.836 | null |
llm-stats:vlmsareblind | llm-stats-vlmsareblind | VLMsAreBlind | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/vlmsareblind?top_n=500 | A vision-language benchmark that probes blind spots and brittle reasoning in multimodal models. | [
"multimodal",
"reasoning",
"vision"
] | [] | multimodal | null | null | unknown | null | null | null | 1 | 4 | 4 | 0.97 | null |
llm-stats:vocalsound | llm-stats-vocalsound | VocalSound | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/vocalsound?top_n=500 | A dataset for improving human vocal sounds recognition, containing over 21,000 crowdsourced recordings of laughter, sighs, coughs, throat clearing, sneezes, and sniffs from 3,365 unique subjects. Used for audio event classification and recognition of human non-speech vocalizations. | [
"audio"
] | [] | audio | null | null | unknown | null | null | null | 1 | 1 | 1 | 0.939 | null |
llm-stats:voicebench-avg | llm-stats-voicebench-avg | VoiceBench Avg | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/voicebench-avg?top_n=500 | VoiceBench is the first benchmark designed to provide a multi-faceted evaluation of LLM-based voice assistants, evaluating capabilities including general knowledge, instruction-following, reasoning, and safety using both synthetic and real spoken instruction data with diverse speaker characteristics and environmental c... | [
"reasoning",
"safety",
"speech_to_text",
"general",
"communication"
] | [] | multimodal | null | null | unknown | null | null | null | 1 | 2 | 2 | 0.901 | null |
llm-stats:vqa-rad | llm-stats-vqa-rad | VQA-Rad | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/vqa-rad?top_n=500 | VQA-RAD (Visual Question Answering in Radiology) is the first manually constructed dataset of medical visual question answering containing 3,515 clinically generated visual questions and answers about radiology images. The dataset includes questions created by clinical trainees on 315 radiology images from MedPix cover... | [
"multimodal",
"image_to_text",
"healthcare",
"vision"
] | [] | multimodal | null | null | unknown | null | null | null | 1 | 1 | 1 | 0.499 | null |
llm-stats:vqav2 | llm-stats-vqav2 | VQAv2 | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/vqav2?top_n=500 | VQAv2 is a balanced Visual Question Answering dataset that addresses language bias by providing complementary images for each question, forcing models to rely on visual understanding rather than language priors. It contains approximately twice the number of image-question pairs compared to the original VQA dataset. | [
"multimodal",
"reasoning",
"image_to_text",
"vision"
] | [] | multimodal | null | null | unknown | null | null | null | 1 | 3 | 3 | 0.809 | null |
llm-stats:vqav2-(test) | llm-stats-vqav2-test | VQAv2 (test) | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/vqav2-%28test%29?top_n=500 | VQA v2.0 (Visual Question Answering v2.0) is a balanced dataset designed to counter language priors in visual question answering. It consists of complementary image pairs where the same question yields different answers, forcing models to rely on visual understanding rather than language bias. The dataset contains 1,10... | [
"multimodal",
"reasoning",
"image_to_text",
"vision"
] | [] | multimodal | null | null | unknown | null | null | null | 1 | 1 | 1 | 0.752 | null |
llm-stats:vqav2-(val) | llm-stats-vqav2-val | VQAv2 (val) | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/vqav2-%28val%29?top_n=500 | VQAv2 is a balanced Visual Question Answering dataset containing open-ended questions about images that require understanding of vision, language, and commonsense knowledge to answer. VQAv2 addresses bias issues from the original VQA dataset by collecting complementary images such that every question is associated with... | [
"multimodal",
"reasoning",
"image_to_text",
"language",
"vision"
] | [] | multimodal | null | null | unknown | null | null | null | 1 | 3 | 3 | 0.716 | null |
opencompass:2383 | opencompass-2383-vrbench | VRBench | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/VRBench | VRBench is the first evaluation benchmark specifically designed to assess multi-step reasoning capabilities over long videos. The benchmark comprises 1,010 long videos with an average duration of 1.6 hours, covering 8 languages and 7 video types. It includes annotations for 9,468 reasoning steps. VRBench是首个专门针对长视频多步推理能... | [
"多模态",
"Multimodal",
"推理",
"Reasoning",
"视频理解",
"Video-Understanding",
"多步推理",
"视频推理",
"多语种",
"多模态模型",
"VLM",
"逻辑推理",
"Video Understanding",
"不支持",
"Unsupported"
] | [] | multimodal | 上海人工智能实验室 | 2025-06-12T00:00:00 | open | https://arxiv.org/abs/2506.10857 | https://github.com/OpenGVLab/VRBench | https://huggingface.co/datasets/OpenGVLab/VRBench | 1 | null | 0 | null | null |
llm-stats:we-math | llm-stats-we-math | We-Math | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/we-math?top_n=500 | We-Math evaluates multimodal models on visual mathematical reasoning, requiring models to understand and solve math problems presented with visual elements such as diagrams, charts, and geometric figures. | [
"math",
"reasoning",
"vision"
] | [] | multimodal | null | null | unknown | null | null | null | 1 | 1 | 1 | 0.89 | null |
llm-stats:web-bench | llm-stats-web-bench | Web Bench | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/web-bench?top_n=500 | Web Bench evaluates agents on realistic web-development engineering tasks, measuring end-to-end implementation in browser-based workflows. | [
"agents",
"coding"
] | [] | text | null | null | unknown | null | null | null | 1 | 2 | 2 | 0.784 | null |
llm-stats:webarena-verified | llm-stats-webarena-verified | WebArena-Verified | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/webarena-verified?top_n=500 | WebArena-Verified evaluates browser agents on realistic web tasks using a verified task set and execution-based grading. | [
"multimodal",
"agents",
"vision"
] | [] | multimodal | null | null | unknown | null | null | null | 1 | 1 | 1 | 0.648 | null |
llm-stats:webdev-arena | llm-stats-webdev-arena | WebDev Arena | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/webdev-arena?top_n=500 | WebDev Arena is a leaderboard for evaluating AI models on web development tasks, including zero-shot generation, complex prompts, and interactive web UI creation. Models are ranked using Elo ratings based on their performance in coding and web development challenges. | [
"reasoning",
"frontend_development",
"code"
] | [] | text | null | null | unknown | null | null | null | 1 | 1 | 1 | 1,588 | null |
opencompass:1982 | opencompass-1982-webui-bench | WebUI-Bench | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/WebUI-Bench | WebUIBench is a comprehensive benchmark designed to evaluate Multimodal Large Language Models (MLLMs) in Web UI-to-code generation tasks across four key capabilities: UI perception, HTML programming, UI-code understanding, and end-to-end transformation. WebUIBench 是一个面向多模态大语言模型（MLLMs）的综合性评测基准，旨在系统评估模型在 Web UI 到代码生成任务中的... | [
"代码",
"Code",
"智能体",
"Agent",
"任务执行",
"Task Execution",
"代码工程",
"不支持",
"Unsupported"
] | [] | null | The Chinese University of HongKong,HongKong SAR,China ,etc. | 2025-06-09T00:00:00 | open | https://arxiv.org/abs/2506.07818 | https://github.com/MAIL-Tele-AI/WebUIBench | https://huggingface.co/datasets/Tele-AI-MAIL/WebUIBench | 1 | null | 0 | null | null |
llm-stats:webvoyager | llm-stats-webvoyager | WebVoyager | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/webvoyager?top_n=500 | WebVoyager evaluates an agent's ability to navigate and complete tasks on real websites by perceiving page screenshots and executing browser actions. | [
"agents",
"vision"
] | [] | multimodal | null | null | unknown | null | null | null | 1 | 1 | 1 | 0.885 | null |
opencompass:1279 | opencompass-1279-whodunitbench | WhodunitBench | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/WhodunitBench | WhodunitBench is used to evaluate large multimodal agent under complex tasks and dynamic scenarios. WhodunitBench用于评估大型多模式代理在复杂任务场景下的动态评估。 | [
"智能体",
"Agent",
"NeurIPS 2024",
"任务执行",
"Task Execution",
"不支持",
"Unsupported"
] | [] | null | The Chinese University of Hong Kong | 2024-09-26T00:00:00 | unknown | null | https://github.com/jun0wanan/WhodunitBench-Murder_Mystery_Games | null | 1 | null | 0 | null | null |
opencompass:504 | opencompass-504-wic | WiC | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/WiC | WiC is a benchmark for the evaluation of context-sensitive word embeddings. WiC is framed as a binary classification task. Each instance in WiC has a target word w, either a verb or a noun, for which two contexts are provided. Each of these contexts triggers a specific meaning of w. The task is to identify if the occur... | [
"语言",
"Language",
"大语言模型",
"LLM",
"语言理解",
"Comprehension",
"开源收录",
"Open-Source",
"不支持",
"Unsupported"
] | [] | null | null | 2018-08-28T00:00:00 | unknown | https://arxiv.org/abs/1808.09121 | null | null | 1 | null | 0 | null | null |
llm-stats:widesearch | llm-stats-widesearch | WideSearch | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/widesearch?top_n=500 | WideSearch is an agentic search benchmark that evaluates models' ability to perform broad, parallel search operations across multiple sources. It tests wide-coverage information retrieval and synthesis capabilities. | [
"reasoning",
"search",
"agents"
] | [] | text | null | null | unknown | null | null | null | 1 | 10 | 10 | 0.819 | null |
model-reports:widesearch | widesearch | WideSearch | model_reports | https://widesearch-seed.github.io/ | Wide-coverage web collection; scored on completeness of an enumerated answer set. | [
"agent"
] | [] | null | null | 2025-08-11T00:00:00 | unknown | null | null | null | 2 | 2 | 2 | 83.9 | percent |
opencompass:1318 | opencompass-1318-wikicontradict | WikiContradict | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/WikiContradict | WikiContradict is a benchmark consisting of 253 high-quality, human-annotated instances designed to assess LLM performance when augmented with retrieved passages containing real-world knowledge conflicts. WikiContradict旨在评估LLM遇到包含真实世界知识冲突的段落检索增强时的性能，由253个高质量的人工注释实例组成。 | [
"推理",
"Reasoning",
"NeurIPS 2024",
"大语言模型",
"LLM",
"事实可靠性",
"Factual Reliability",
"逻辑推理",
"不支持",
"Unsupported"
] | [
"English"
] | null | IBM Research | 2024-06-19T00:00:00 | unknown | https://arxiv.org/abs/2406.13805 | null | https://huggingface.co/datasets/ibm/Wikipedia_contradict_benchmark | 1 | null | 0 | null | null |
opencompass:1129 | opencompass-1129-wikisql | WikiSQL | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/WikiSQL | WikiSQL is a dataset of 80654 hand-annotated examples of questions and SQL queries distributed across 24241 tables from Wikipedia that is an order of magnitude larger than comparable datasets. WikiSQL 是一个包含 80,654 个手动标注示例的问题和 SQL 查询的数据集，分布在来自维基百科的 24,241 个表格中。 | [
"其他",
"Other",
"大语言模型",
"LLM",
"代码工程",
"Code",
"不支持",
"Unsupported"
] | [] | null | Salesforce Research | 2017-11-09T00:00:00 | unknown | https://arxiv.org/pdf/1709.00103v7 | https://github.com/salesforce/WikiSQL | null | 1 | null | 0 | null | null |
llm-stats:wild-bench | llm-stats-wild-bench | Wild Bench | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/wild-bench?top_n=500 | WildBench is an automated evaluation framework that benchmarks large language models using 1,024 challenging, real-world tasks selected from over one million human-chatbot conversation logs. It introduces two evaluation metrics (WB-Reward and WB-Score) that achieve high correlation with human preferences and uses task-... | [
"reasoning",
"general",
"communication"
] | [] | text | null | null | unknown | null | null | null | 1 | 8 | 8 | 0.685 | null |
opencompass:1555 | opencompass-1555-wildbench | WildBench | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/WildBench | Weintroduce WildBench, an automated evaluation framework designed to bench-mark large language models (LLMs) using challenging, real-world user queries, which consists of 1,024 tasks carefully selected from over one million human-chatbot conversation logs. WildBench推出自动评估框架和数据集，基于真实用户难题评测大语言模型，包含从逾百万人机对话日志中精选的1,024个任务样... | [
"理解",
"Understanding",
"大语言模型",
"LLM",
"语言生成",
"Generation",
"指令遵循",
"Instruction Following",
"不支持",
"Unsupported"
] | [] | null | Allen Institute for AI, University of Washington | 2024-06-07T00:00:00 | restricted | https://arxiv.org/abs/2406.04770 | https://github.com/allenai/WildBench | https://huggingface.co/datasets/allenai/WildChat | 1 | null | 0 | null | null |
llm-stats:wildclawbench | llm-stats-wildclawbench | WildClawBench | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/wildclawbench?top_n=500 | WildClawBench is an agentic coding benchmark from InternLM/Claw-Eval that reports overall model performance on real-world tool-using development tasks. | [
"agents",
"coding"
] | [] | text | null | null | unknown | null | null | null | 1 | 5 | 5 | 0.628 | null |
opencompass:2445 | opencompass-2445-wildclawbench | WildClawBench | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/WildClawBench | WildClawBench evaluates AI agents across six task categories: productivity flow, code intelligence, social interaction, search & retrieval, creative synthesis, and safety alignment. It comprises 60 hand-authored tasks inside a live environment with real tools — browser, bash, file system, etc. WildClawBench 从六个任务维度评测 A... | [
"智能体",
"Agent",
"代码",
"Code",
"创作",
"Creation",
"Agent Evaluation",
"In-the-Wild Tasks",
"Multimodal & Tool-Use Reasoning",
"代码工程",
"语言生成",
"Generation",
"任务执行",
"Task Execution",
"跨模态推理",
"Cross-modal Reasoning",
"不支持",
"Unsupported"
] | [
"English",
"Chinese"
] | null | 上海人工智能实验室 | 2026-04-07T00:00:00 | open | null | https://github.com/InternLM/WildClawBench | https://huggingface.co/datasets/internlm/WildClawBench | 1 | null | 0 | null | null |
llm-stats:winogrande | llm-stats-winogrande | Winogrande | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/winogrande?top_n=500 | WinoGrande: An Adversarial Winograd Schema Challenge at Scale. A large-scale dataset of 44,000 pronoun resolution problems designed to test machine commonsense reasoning. Uses adversarial filtering to reduce spurious biases and provides a more robust evaluation of whether AI systems truly understand commonsense or expl... | [
"reasoning",
"language"
] | [] | text | Allen Institute for Artificial Intelligence | 2019-11-21T00:00:00 | unknown | https://arxiv.org/pdf/1907.10641 | https://github.com/allenai/winogrande | null | 1 | 22 | 22 | 0.875 | null |
opencompass:1109 | opencompass-1109-winogrande | WinoGrande | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/WinoGrande | WINOGRANDE is a large-scale dataset of 44k problems, inspired by the original WSC design, but adjusted to improve both the scale and the hardness of the dataset. WINOGRANDE 包含 44,000 个问题，受到 WSC 设计的启发，但进行了调整，以提高数据集的规模和难度。 | [
"推理",
"Reasoning",
"大语言模型",
"LLM",
"逻辑推理",
"开源收录",
"Open-Source",
"支持",
"Supported"
] | [] | null | Allen Institute for Artificial Intelligence | 2019-11-21T00:00:00 | unknown | https://arxiv.org/pdf/1907.10641 | https://github.com/allenai/winogrande | null | 1 | null | 0 | null | null |
llm-stats:wmdp | llm-stats-wmdp | WMDP | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/wmdp?top_n=500 | Weapons of Mass Destruction (WMDP) is a multiple-choice benchmark on dual-use biology, chemistry, and cyber knowledge. It measures a model's capacity to enable malicious actors to design, synthesize, acquire, or use chemical, biological, radiological, or nuclear (CBRN) weapons. | [
"safety",
"healthcare",
"biology",
"chemistry"
] | [] | text | null | null | unknown | null | null | null | 1 | 1 | 1 | 0.84 | null |
llm-stats:wmt23 | llm-stats-wmt23 | WMT23 | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/wmt23?top_n=500 | The Eighth Conference on Machine Translation (WMT23) benchmark evaluating machine translation systems across 8 language pairs (14 translation directions) including general, biomedical, literary, and low-resource language translation tasks. Features specialized shared tasks for quality estimation, metrics evaluation, si... | [
"language",
"healthcare"
] | [] | text | null | null | unknown | null | null | null | 1 | 4 | 4 | 0.751 | null |
llm-stats:wmt24++ | llm-stats-wmt24 | WMT24++ | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/wmt24%2B%2B?top_n=500 | WMT24++ is a comprehensive multilingual machine translation benchmark that expands the WMT24 dataset to cover 55 languages and dialects. It includes human-written references and post-edits across four domains (literary, news, social, and speech) to evaluate machine translation systems and large language models across d... | [
"language"
] | [] | text | null | null | unknown | null | null | null | 1 | 23 | 23 | 0.8667 | null |
llm-stats:workspace-bench | llm-stats-workspace-bench | Workspace Bench | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/workspace-bench?top_n=500 | Workspace Bench evaluates AI agents on high-economic-value workplace tasks that span multi-step planning, file processing, and tool use across realistic office and productivity workflows. | [
"reasoning",
"general",
"agents"
] | [] | text | null | null | unknown | null | null | null | 1 | 3 | 3 | 0.677 | null |
model-reports:workspacebench | workspacebench | WorkspaceBench | model_reports | https://arxiv.org/abs/2605.03596 | Large-scale file-workspace tasks; the environment and file dependencies are part of the measurement. | [
"professional"
] | [] | null | null | 2026-05-05T00:00:00 | unknown | https://arxiv.org/abs/2605.03596 | null | null | 1 | 1 | 1 | 60.2 | percent |
llm-stats:worldbench | llm-stats-worldbench | WorldBench | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/worldbench?top_n=500 | WorldBench evaluates real-world visual knowledge and understanding across diverse everyday scenes. | [
"multimodal",
"knowledge",
"vision"
] | [] | multimodal | null | null | unknown | null | null | null | 1 | 2 | 2 | 0.676 | null |
opencompass:1938 | opencompass-1938-worldgenbench | WorldGenBench | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/WorldGenBench | [
"多模态",
"Multimodal",
"推理",
"Reasoning",
"知识",
"Knowledge",
"多模态模型",
"VLM",
"视觉生成",
"Visual Generation",
"知识储备",
"不支持",
"Unsupported"
] | [] | multimodal | null | 2025-06-16T00:00:00 | unknown | https://arxiv.org/abs/2505.01490 | null | https://huggingface.co/datasets/worldrl/WorldGenBench | 1 | null | 0 | null | null | |
opencompass:1735 | opencompass-1735-worldscore | WorldScore | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/WorldScore | WorldScore benchmark is the first unified benchmark for world generation. WorldScore基准测试，这是首个用于世界生成的统一基准测试。 | [
"创作",
"Creation",
"其他",
"Other",
"视频生成",
"多模态模型",
"VLM",
"视觉生成",
"Visual Generation",
"不支持",
"Unsupported"
] | [] | null | Stanford University | 2025-04-01T00:00:00 | open | https://arxiv.org/abs/2504.00983 | https://github.com/haoyi-duan/WorldScore | https://huggingface.co/datasets/Howieeeee/WorldScore | 1 | null | 0 | null | null |
llm-stats:worldvqa | llm-stats-worldvqa | WorldVQA | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/worldvqa?top_n=500 | WorldVQA is a benchmark designed to evaluate atomic vision-centric world knowledge. It assesses models' ability to understand and reason about visual elements representing real-world knowledge. | [
"multimodal",
"reasoning",
"vision"
] | [] | multimodal | null | null | unknown | null | null | null | 1 | 5 | 5 | 0.611 | null |
llm-stats:writingbench | llm-stats-writingbench | WritingBench | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/writingbench?top_n=500 | A comprehensive benchmark for evaluating large language models' generative writing capabilities across 6 core writing domains (Academic & Engineering, Finance & Business, Politics & Law, Literature & Art, Education, Advertising & Marketing) and 100 subdomains. Contains 1,239 queries with a query-dependent evaluation fr... | [
"legal",
"finance",
"communication",
"creativity",
"writing"
] | [] | text | null | null | unknown | null | null | null | 1 | 15 | 15 | 0.883 | null |
opencompass:1701 | opencompass-1701-writingbench | WritingBench | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/WritingBench | WritingBench: A Comprehensive Benchmark for Generative Writing WritingBench: A Comprehensive Benchmark for Generative Writing | [
"长文本",
"Long-Context",
"创作",
"Creation",
"大语言模型",
"LLM",
"长上下文",
"Long Context",
"语言生成",
"Generation",
"不支持",
"Unsupported"
] | [] | null | Alibaba Group; Renmin University of China; Shanghai Jiao Tong University | 2025-03-07T00:00:00 | unknown | https://arxiv.org/pdf/2503.05244 | https://github.com/X-PLUG/WritingBench | null | 1 | null | 0 | null | null |
opencompass:507 | opencompass-507-wsc | WSC | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/WSC | WSC is a pronoun disambiguation task, which requires to determine which noun the pronoun refers to according to the context. WSC是一个代词消歧任务，要求根据上下文判断代词指代的是哪个名词。 | [
"语言",
"Language",
"大语言模型",
"LLM",
"语言理解",
"Comprehension",
"开源收录",
"Open-Source",
"不支持",
"Unsupported"
] | [] | null | null | null | unknown | null | null | null | 1 | null | 0 | null | null |
opencompass:1072 | opencompass-1072-xcodeeval | xCodeEval | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/xCodeEval | xCodeEval is the largest executable multilingual multitask benchmark to date consisting of 25M document-level coding examples (16.5B tokens) from about 7.5K unique problems covering up to 11 programming languages with execution-level parallelism. xCodeEval 是迄今为止最大的可执行多语言多任务基准，包含 2500 万个文档级编码示例（165 亿个标记），来自约 7500 个独特问题，... | [
"代码",
"Code",
"ACL 2024",
"大语言模型",
"LLM",
"代码工程",
"不支持",
"Unsupported"
] | [
"Multilingual"
] | null | NTU-NLP | 2023-11-06T00:00:00 | open | https://arxiv.org/pdf/2303.03004 | https://github.com/ntunlp/xCodeEval | https://huggingface.co/datasets/NTU-NLP-sg/xCodeEval | 1 | null | 0 | null | null |
llm-stats:xdailybench | llm-stats-xdailybench | xDailyBench | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/xdailybench?top_n=500 | xDailyBench evaluates AI agents on white-collar office work, covering everyday professional tasks such as document handling, consultation, and multi-step productivity workflows. | [
"reasoning",
"general",
"agents"
] | [] | text | null | null | unknown | null | null | null | 1 | 2 | 2 | 0.61 | null |
llm-stats:xlsum-english | llm-stats-xlsum-english | XLSum English | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/xlsum-english?top_n=500 | Large-scale multilingual abstractive summarization dataset comprising 1 million professionally annotated article-summary pairs from BBC, covering 44 languages. XL-Sum is highly abstractive, concise, and of high quality, designed to encourage research on multilingual abstractive summarization tasks. | [
"summarization",
"language"
] | [] | text | null | null | unknown | null | null | null | 1 | 1 | 1 | 0.3161 | null |
llm-stats:xstest | llm-stats-xstest | XSTest | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/xstest?top_n=500 | XSTest is a test suite designed to identify exaggerated safety behaviours in large language models. It comprises 450 prompts: 250 safe prompts across ten prompt types that well-calibrated models should not refuse to comply with, and 200 unsafe prompts as contrasts that models should refuse. The benchmark systematically... | [
"safety"
] | [] | text | null | null | unknown | null | null | null | 1 | 4 | 4 | 0.988 | null |
opencompass:521 | opencompass-521-xsum | XSum | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/XSum | XSum is a single-document summarization task which does not favor extractive strategies and calls for an abstractive modeling approach. The idea is to create a short, one-sentence news summary answering the question “What is the article about?”. The dataset collects real-world, large scale data by harvesting online art... | [
"理解",
"Understanding",
"大语言模型",
"LLM",
"语言理解",
"Comprehension",
"开源收录",
"Open-Source",
"不支持",
"Unsupported"
] | [] | null | null | 2018-08-27T00:00:00 | unknown | https://arxiv.org/abs/1808.08745 | https://github.com/EdinburghNLP/XSum | null | 1 | null | 0 | null | null |
opencompass:1784 | opencompass-1784-xverify | xVerify | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/xVerify | xVerify, an efficient answer verifier for reasoning model evaluations. xVerify demonstrates strong capability in equivalence judgment, enabling it to effectively determine whether the answers produced by reasoning models are equivalent to reference answers across various types of objective questions xVerify，这是一种用于推理模型评... | [
"推理",
"Reasoning",
"大语言模型",
"LLM",
"逻辑推理",
"不支持",
"Unsupported"
] | [
"English",
"Chinese"
] | null | Peking University，Research Institute of China Telecom，etc. | 2025-04-14T00:00:00 | unknown | https://arxiv.org/abs/2504.10481 | https://github.com/IAAR-Shanghai/xVerify | null | 1 | null | 0 | null | null |
llm-stats:yc-bench | llm-stats-yc-bench | YC-Bench | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/yc-bench?top_n=500 | YC-Bench evaluates agents on long-horizon, open-ended business and investment decision-making. The reported metric is the final assets (fund value, in US dollars) accumulated by the agent over the course of the simulation. | [
"finance",
"agents"
] | [] | text | null | null | unknown | null | null | null | 1 | 1 | 1 | 2,100,000 | null |
opencompass:1017 | opencompass-1017-yue-benchmark | Yue_Benchmark | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/Yue_Benchmark | The benchmarks introduced for evaluating large language models (LLMs) on Cantonese include Yue-TruthfulQA, Yue-GSM8K, Yue-ARC-C, Yue-MMLU, and Yue-TRANS. Each of these benchmarks focuses on different aspects of language understanding and generation in Cantonese, offering a comprehensive means of assessing the capabilit... | [
"语言",
"Language",
"大语言模型",
"LLM",
"语言理解",
"Comprehension",
"不支持",
"Unsupported"
] | [
"English"
] | null | null | 2024-08-31T00:00:00 | open | https://arxiv.org/abs/2408.16756 | https://github.com/jiangjyjy/Yue-Benchmark | https://huggingface.co/datasets/BillBao/Yue-Benchmark | 1 | null | 0 | null | null |
llm-stats:zclawbench | llm-stats-zclawbench | ZClawBench | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/zclawbench?top_n=500 | ZClawBench evaluates Claw-style agent task execution quality, measuring a model's ability to autonomously complete complex multi-step coding tasks in real-world environments. | [
"agents",
"code"
] | [] | text | null | null | unknown | null | null | null | 1 | 4 | 4 | 0.643 | null |
llm-stats:zebralogic | llm-stats-zebralogic | ZebraLogic | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/zebralogic?top_n=500 | ZebraLogic is an evaluation framework for assessing large language models' logical reasoning capabilities through logic grid puzzles derived from constraint satisfaction problems (CSPs). The benchmark consists of 1,000 programmatically generated puzzles with controllable and quantifiable complexity, revealing a 'curse ... | [
"reasoning"
] | [] | text | null | null | unknown | null | null | null | 1 | 8 | 8 | 0.973 | null |
llm-stats:zerobench | llm-stats-zerobench | ZEROBench | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/zerobench?top_n=500 | ZEROBench is a challenging vision benchmark designed to test models on zero-shot visual understanding tasks. | [
"multimodal",
"reasoning",
"vision"
] | [] | image | null | null | unknown | null | null | null | 1 | 9 | 9 | 0.41 | null |
opencompass:1532 | opencompass-1532-zerobench | ZeroBench | opencompass_hub | https://hub.opencompass.org.cn/dataset-detail/ZeroBench | ZeroBench is a challenging visual reasoning benchmark for LMMs. It consists of a main set of 100 high-quality, manually curated questions covering numerous domains, reasoning types and image type. Questions have been designed and calibrated to be beyond the capabilities of current frontier models. ZeroBench 是针对多模态模型（LM... | [
"多模态",
"Multimodal",
"多模态模型",
"VLM",
"跨模态推理",
"Cross-modal Reasoning",
"不支持",
"Unsupported"
] | [] | multimodal | University of Cambridge, University of Alberta,etc. | 2025-02-13T00:00:00 | restricted | https://arxiv.org/abs/2502.09696 | https://github.com/jonathan-roberts1/zerobench | https://huggingface.co/datasets/jonathan-roberts1/zerobench | 1 | null | 0 | null | null |
llm-stats:zerobench-sub | llm-stats-zerobench-sub | ZEROBench-Sub | llm_stats | https://api.zeroeval.com/leaderboard/benchmarks/zerobench-sub?top_n=500 | ZEROBench-Sub is a subset of the ZEROBench benchmark. | [
"multimodal",
"reasoning",
"vision"
] | [] | image | null | null | unknown | null | null | null | 1 | 5 | 5 | 0.362 | null |
artificial-analysis:tau2-bench-telecom | artificial-analysis-tau2-bench-telecom | τ²-Bench Telecom | artificial_analysis | https://artificialanalysis.ai/evaluations/tau2-bench | Agentic tool use | [
"agentic",
"tool-use"
] | [] | null | null | 2025-06-12T00:00:00 | unknown | null | null | null | 1 | 440 | 440 | 0.991228 | null |
artificial-analysis:tau3-banking | artificial-analysis-tau3-banking | τ³-Banking | artificial_analysis | https://artificialanalysis.ai/evaluations/tau3-banking | Agentic tool use | [
"intelligence-index",
"agentic",
"tool-use"
] | [] | null | null | null | unknown | null | null | null | 1 | 178 | 178 | 0.513402 | null |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.