---
title: BenchHub Leaderboards
emoji: 🏆
colorFrom: purple
colorTo: indigo
sdk: static
app_file: index.html
pinned: true
hf_oauth: false
short_description: Reproducible LLM/vision/audio leaderboards. Submit free
tags:
  - leaderboard
  - benchmark
  - evaluation
  - model-evaluation
  - reproducibility
  - llm
  - text-generation
  - reasoning
  - mmlu
  - computer-vision
  - image-classification
  - object-detection
  - image-segmentation
  - depth-estimation
  - optical-flow
  - stereo
  - audio-classification
  - automatic-speech-recognition
  - speech-recognition
  - question-answering
  - named-entity-recognition
  - token-classification
datasets:
  - AI-Lab-Makerere/beans
  - Aaryan333/fer2013_train_publicTest_privateTest
  - Bingsu/Cat_and_Dog
  - Bingsu/Human_Action_Recognition
  - Falah/Alzheimer_MRI
  - Hemg/Brain-Tumor-MRI-Dataset
  - Joker5800/Tobacco_Leaf_Diseases
  - Marxulia/asl_sign_languages_alphabets_v03
  - Multimodal-Fatima/Caltech101_not_background_test
  - Multimodal-Fatima/FGVC_Aircraft_test
  - Nattakarn/fruit-and-vegetable-image-recognition
  - POSE-Lab/IndustryShapes
  - Rishi210904/Crop_Weed_classification
  - Rowan/hellaswag
  - Spitblaze/Corn_or_Maize_Leaf_Disease_Dataset
  - TIGER-Lab/MMLU-Pro
  - Teklia/IAM-line
  - TheKernel01/140k-Real-and-Fake-Faces
  - abisee/cnn_dailymail
  - ahmed-ai/skin-lesions-classification-dataset
  - allenai/ai2_arc
  - allenai/openbookqa
  - allenai/qasc
  - allenai/sciq
  - allenai/winogrande
  - amaye15/stanford-dogs
  - anthony2261/paddy-disease-classification
  - aps/super_glue
  - ashraq/esc50
  - blanchon/EuroSAT_RGB
  - blanchon/PatternNet
  - blanchon/UC_Merced
  - cais/mmlu
  - cornell-movie-review-data/rotten_tomatoes
  - dair-ai/emotion
  - deanngkl/raf-db-7emotions
  - deepmind/aqua_rat
  - detection-datasets/fashionpedia
  - dpdl-benchmark/oxford_flowers102
  - efekankavalci/CUB_200_2011
  - ehovy/race
  - ethz/food101
  - facebook/anli
  - fancyzhx/ag_news
  - flaviagiammarino/vqa-rad
  - google-research-datasets/conceptual_captions
  - google/boolq
  - google/speech_commands
  - hails/agieval-logiqa-en
  - hails/agieval-lsat-lr
  - jagennath-hari/nyuv2
  - jonathan-roberts1/GID
  - jonathan-roberts1/NWPU-RESISC45
  - jonathan-roberts1/Optimal-31
  - jonathan-roberts1/SIRI-WHU
  - jonathan-roberts1/Satellite-Images-of-Hurricane-Damage
  - jonathan-roberts1/USTC_SmokeRS
  - keremberke/blood-cell-object-detection
  - keremberke/csgo-object-detection
  - keremberke/excavator-detector
  - keremberke/forklift-object-detection
  - keremberke/hard-hat-detection
  - keremberke/license-plate-object-detection
  - keremberke/plane-detection
  - keremberke/valorant-object-detection
  - lighteval/piqa
  - lighteval/siqa
  - liventruth/PSAS-Acoustic-Verification
  - naufalso/carla_hd
  - nelorth/oxford-flowers
  - openai/gsm8k
  - openlifescienceai/medmcqa
  - openslr/librispeech_asr
  - prithivMLmods/Face-Age-10K
  - prs-eth/ZuriPano
  - qingyuyang/Fetal_Planes_DB
  - rafaelpadilla/coco2017
  - rajpurkar/squad
  - rotemvahava/yoga-poses-107
  - sartajbhuvaji/Brain-Tumor-Classification
  - stanfordnlp/imdb
  - stanfordnlp/snli
  - stanfordnlp/sst2
  - stereo-dataset/stereo-dataset
  - tanganke/dtd
  - tanganke/eurosat
  - tanganke/gtsrb
  - tanganke/kmnist
  - tanganke/resisc45
  - tanganke/stanford_cars
  - tanganke/stl10
  - tau/commonsense_qa
  - timm/mini-imagenet
  - timm/oxford-iiit-pet
  - timm/resisc45
  - trpakov/chest-xray-classification
  - truthfulqa/truthful_qa
  - uoft-cs/cifar10
  - uoft-cs/cifar100
  - wmt/wmt14
  - xbgoose/ravdess
  - ylecun/mnist
  - zalando-datasets/fashion_mnist
  - zh-plus/tiny-imagenet
---

# 🏆 BenchHub Leaderboards

A read-only mirror of **public** leaderboard standings from
**[runbenchhub.com](https://runbenchhub.com)** — an open, multi-modal model
benchmarking platform. Browse the standings here; **run your own model and
submit on BenchHub** (free).

## What's benchmarked
Live boards across a growing set of domains, each with real models scored on
the same eval set:

- **LLM / Reasoning** — MMLU (14k MCQ, 57 subjects), HellaSwag, ARC-Challenge,
  GSM8K — one **pinned prompt**, scored by exact match, zero-shot, so every
  model is evaluated identically
- **Vision** — Image Classification, Semantic Segmentation (mIoU), Object
  Detection (mAP), Optical Flow, Monocular & **Stereo** Depth Estimation,
  Point Tracking, Image Captioning
- **Audio** — Audio Classification (ESC-50), Automatic Speech Recognition (WER)
- **NLP** — Text Classification, Extractive Question Answering (SQuAD),
  Named-Entity Recognition (CoNLL-2003), Visual Question Answering, Translation

## Submit your model
Every **"Submit"** button here deep-links to that leaderboard's submission page
on BenchHub. Submitting is free — sign in with **GitHub, Google, or 🤗 Hugging
Face**, run the one-line client on your predictions, and see where you rank.

> This Space holds **no** ground-truth samples, predictions, or submission UI —
> it reads a derived standings dataset (`HF_RESULTS_REPO`) that BenchHub
> publishes. The full interactive experience (per-sample explorer, GT
> visualizations, model comparison) lives on
> **[runbenchhub.com](https://runbenchhub.com)**.