Qwen2.5 VL 72B is the stronger all-rounder.
Scores updated · 22 tests both models report · How we compare
Where each one wins
Tests won in each of the two areas where both have results. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
Agents rests on a single test.
What it costs
Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where Qwen2.5 VL 72B pulls ahead
Every test, side by side
All 22 tests both models report. The winning score is in its model's colour; marks a score checked independently.
Images and chartsQwen2.5 VL 72B
Other results16 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- MATH-VisionQwen2.5 VL 72B by 7.730.438.1+7.7
- MathVision (full) (Pass@1)Qwen2.5 VL 72B by 7.730.438.1+7.7
- MMVet (Pass@1)Qwen2.5 VL 72B by 7.169.176.2+7.1
- OCRBench (Acc)Qwen2.5 VL 72B by 70 rating points815885+70 rating
- MMStar (Acc)Qwen2.5 VL 72B by 6.164.770.8+6.1
- AI2DGPT-4o by 5.894.288.4+5.8
- MMBench-EN-v1.1 (Acc)Qwen2.5 VL 72B by 5.283.188.3+5.2
- MMVet (Acc)Qwen2.5 VL 72B by 4.969.174+4.9
- EgoSchemaQwen2.5 VL 72B by 472.276.2+4
- MMLongBench-DOC (Acc)GPT-4o by 442.838.8+4
- ChartQAQwen2.5 VL 72B by 3.885.789.5+3.8
- DocVQAQwen2.5 VL 72B by 3.692.896.4+3.6
- MMMU (val) (Pass@1)GPT-4o by 3.369.165.8+3.3
- VideoMME (w sub.)Qwen2.5 VL 72B by 1.977.279.1+1.9
- RealWorldQAGPT-4o by 1.575.473.9+1.5
- RealWorldQA (Acc)tie75.475.7tie
Questions people ask
Which is better, GPT-4o or Qwen2.5 VL 72B?
Qwen2.5 VL 72B wins all two areas where both have results: agents and images and charts. GPT-4o wins none.
How do you compare the two?
We use the 22 benchmark tests both models have published scores on. The verdict counts the 6 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 16 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.