ERNIE 5.0 Technical Report

Benchmark Model Rank Results
agents-on-benchERNIE 5.0–Accuracy: 78.79
agents-on-bfclv4ERNIE 5.0–Accuracy: 66.47
audio-classification-on-vocalsoundERNIE 5.0–Accuracy: 90.73
audio-understanding-reasoning-on-mmsuERNIE 5.0–Accuracy: 84.68
chart-question-answering-on-chartqaERNIE 5.0–1:1 Accuracy: 87.8
deep-research-agents-on-browsecomp-zhERNIE 5.0–Accuracy: 64.71
instruction-following-on-multi-ifERNIE 5.0–Accuracy: 85.56
logic-puzzle-solving-on-zebralogicERNIE 5.0–Accuracy: 96.5
mathematical-reasoning-on-mathverseERNIE 5.0–Accuracy: 85.13
multi-task-language-understanding-on-mmlu-proERNIE 5.0–Accuracy: 83.8
multi-task-language-understanding-on-mmmluERNIE 5.0-Base–Accuracy: 78.94
multimodal-reasoning-on-mathvisionERNIE 5.0–Accuracy: 74.34
multimodal-reasoning-on-mathvistaERNIE 5.0–Accuracy: 84.8
multimodal-reasoning-on-mmmu-proERNIE 5.0–Accuracy: 68.63
multimodal-reasoning-on-visulogicERNIE 5.0–Accuracy: 32.0
multiple-choice-question-answering-mcqa-on-gpqa-diamondERNIE 5.0–Accuracy: 86.36
optical-character-recognition-ocr-on-ocrbenchERNIE 5.0–Score: 878.0
speech-recognition-on-librispeech-test-otherERNIE 5.0–Word Error Rate (WER): 2.61
text-to-speech-synthesis-on-seed-tts-test-enERNIE 5.0–WER: 1.54
video-question-answering-on-mmvuERNIE 5.0–Accuracy: 72.24
video-question-answering-on-video-mmmuERNIE 5.0–Accuracy: 81.11
vision-language-understanding-on-mmstarERNIE 5.0–Accuracy: 75.54
visual-question-answering-vqa-on-simplevqaERNIE 5.0–Accuracy: 67.64