Qwen 2.5 VL 72B Instruct — benchmark results

Alibaba Qwen 2.5 VL 72B vision-language instruction-tuned checkpoint. Provider: Alibaba. Released 2025-01-26. Access: Open.

Unified ELO 1575 ± 16, rank #604 of 1839 rated models, from 75 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
LLM Stats (DocVQA)96.4Score (%)100
StickToYourRole83.6Cardinal Score100
TableVista55.8Avg. (self-reported)96.4
Open Portuguese LLM - Hate Speech76.96F1 (%)93.9
KOFFVQA - Korean OCR95Score (%)93.8
LLM Stats (ChartQA)89.5Score (%)91.3
CapArena-Auto35.3CapArena-Auto Score (avg)90.5
CapArena-Auto vs CogVLM-19B49CapArena-Auto Score (vs CogVLM-19B)90.5
CapArena-Auto vs GPT-4o-1CapArena-Auto Score (vs GPT-4o)90.5
LLM Stats (EgoSchema)76.2Score (%)87.5
LLM Stats (MMBench)88Score (%)87.5
KOFFVQA - Recognition90Score (%)87

Interactive version: theaggregate.ai/model?slug=qwen-2-5-vl-72b-instruct · How It Works · Data refreshed daily, snapshot 2026-08-05.