Codesota · Models · Qwen2.5-VL 72BAlibaba5 results · 5 benchmarks
Model card

Qwen2.5-VL 72B.

Alibabaopen-source72B paramsVision-Language ModelApache 2.0
§ 01 · Benchmarks

Every benchmark Qwen2.5-VL 72B has a recorded score for.

#BenchmarkArea · TaskMetricValueRankDateSource
01MMBenchMultimodal · Visual Question Answeringaccuracy90.5%#1/82025-02-19source ↗
02TextVQAMultimodal · Visual Question Answeringaccuracy85.5%#1/92025-02-19source ↗
03MME-VideoOCRComputer Vision · General OCR Capabilitiestotal-accuracy69.0%#2/6source ↗
04ThaiOCRBenchComputer Vision · Optical Character Recognitionted-score0.7%#5/5source ↗
05MMMUMultimodal · Visual Question Answeringaccuracy70.2%#10/182025-02-19source ↗
Rank column shows this model’s position vs all other models scored on the same benchmark + metric (competitors after the slash). #1 in red means current SOTA. Sorted by rank, then newest result.
§ 02 · Strengths by area

Where Qwen2.5-VL 72B actually performs.

Computer Vision
2
benchmarks
avg rank #3.5
Multimodal
3
benchmarks
avg rank #4.0
§ 03 · Papers

1 paper with results for Qwen2.5-VL 72B.

  1. 2025-02-19· Multimodal· 3 results

    Qwen2.5-VL Technical Report

§ 04 · Related models

Other Alibaba models scored on Codesota.

Qwen2-VL 72B
4 results
Qwen2.5-72B-Instruct
72B params · 4 results
Qwen2.5-Coder 32B
32B params · 4 results
GOT-OCR2.0
3 results
Qwen 3 72B
72B params · 2 results
Qwen2.5-VL 32B
2 results
Qwen 3 14B
14B params · 1 result
Qwen2-VL 7B
7B params · 1 result
§ 05 · Sources & freshness

Where these numbers come from.

arxiv
3
results
alphaxiv-leaderboard
2
results
3 of 5 rows marked verified.