Qwen2.5 VL 72B
Qwen2.5 VL 72B is behind the leaders in multimodal tasks. Too few results yet to rate reasoning, coding, agentic tasks, safety, long context, math, multilingual tasks, instruction following, or factuality.
Capability profile
Bars show the model's median result as a share of the leading model's, per capability. Select a row to see its results.
Too few results yet to rate Reasoning, Coding, Agentic, Safety, Long Context, Math, Multilingual, Instruction Following or Factuality.
Price
No current price is tracked for this model. See the rate card
Evidence
61results on50benchmarks
- 21 vendor-reported
- 40 cross-referenced
From 5 sources · latest Oct 7, 2026 · How verification works
Qwen2.5 VL 72B benchmark results
61 results on 50 benchmarks, grouped by capability. Each bar is the result as a share of the capability leader's; every result links to its source.
35.0% behind the leader4 of 6 ranked benchmarks measured
- 88.50Oct 7, 2026
- MathVista74.80Oct 7, 2026MathVista-Mini
- 70.20Oct 7, 2026
- 51.10Oct 7, 2026
0 of 7 ranked benchmarks measured
- OSWorld-Verified8.83Oct 7, 2026OSWorld
More results
Benchmarks outside the capability baskets. They are not ranked against a leader.
- ScreenSpot-Pro (No tools)43.60Oct 7, 2026ScreenSpot Pro
- 87.10Oct 7, 2026
- OCRBench v2_en61.50Oct 7, 2026OCRBench-V2 (en)
- 70.40Oct 7, 2026
- MMVet (Pass@1)76.19Oct 7, 2026MMVet
- 2.02Oct 7, 2026
Show 43 more resultsHide 43 results
- A-Bench_VAL79.22Sep 2, 2026
- 88.40Oct 7, 2026
- AI2D83.84Sep 2, 2026AI2D_TEST
- 79.80Oct 7, 2026
- 73.73Sep 2, 2026
- 89.50Oct 7, 2026
- 45.30Sep 2, 2026
- 96.40Oct 7, 2026
- 95.75Sep 2, 2026
- 76.20Oct 7, 2026
- 55.16Oct 7, 2026
- 49.90Sep 2, 2026
- 47.30Oct 7, 2026
- 38.10Oct 7, 2026
- 39.34Sep 2, 2026
- MATH-Vision38.10Jun 5, 2026MATH-Vision (Pass@1)
- 55.18Sep 2, 2026
- 38.10Jun 5, 2026
- 88.00Oct 7, 2026
- 88.30Jun 5, 2026
- 38.80Jun 5, 2026
- MMMU (val) (Pass@1)65.78Sep 2, 2026MMMU_VAL
- 74.80Jun 5, 2026
- 70.80Jun 5, 2026
- MMT-Bench_VAL69.49Sep 2, 2026
- 74.00Jun 5, 2026
- MMVet (Pass@1)75.69Sep 2, 2026MMVet
- MMVU62.90Jun 5, 2026MMVU (Pass@1)
- MTVQA_TEST31.48Sep 2, 2026
- 885.00Jun 5, 2026
- OCRVQA_TEST66.80Sep 2, 2026
- 83.35Sep 2, 2026
- Q-Bench1_VAL79.93Sep 2, 2026
- 73.86Sep 2, 2026
- 75.70Jun 5, 2026
- ScienceQA_TEST92.51Sep 2, 2026
- ScienceQA_VAL91.32Sep 2, 2026
- 43.60Jun 5, 2026
- 78.34Sep 2, 2026
- 73.25Sep 2, 2026
- 83.26Sep 2, 2026
- VideoMME (w sub.)79.10Jun 5, 2026Video-MME (w/ sub.)
- VideoMMMU60.20Jun 5, 2026VideoMMMU (Pass@1)
Qwen2.5 VL 72B: common questions
Who makes Qwen2.5 VL 72B?
Qwen2.5 VL 72B is made by Alibaba.
When was Qwen2.5 VL 72B released?
Qwen2.5 VL 72B was released on Jan 26, 2025, according to Alibaba's own announcement.
What is Qwen2.5 VL 72B good at?
Qwen2.5 VL 72B is behind the leaders in multimodal tasks. Too few results yet to rate reasoning, coding, agentic tasks, safety, long context, math, multilingual tasks, instruction following, or factuality.
How many benchmarks has Qwen2.5 VL 72B been tested on?
We track 61 results for Qwen2.5 VL 72B on 50 benchmarks from 5 sources. The latest was recorded on Oct 7, 2026.
About this record
Where Qwen2.5 VL 72B's numbers come from, and every name it appears under.
- Tracked since
- Jun 5, 2026
- Newest source mention
- Aug 5, 2026
Where the results come from
Verification: 61 scores · 0 independently verified · 40 vendor cross-reference · 21 vendor-reported. How these tiers are assigned
From 5 sources on 2 sites. Hugging Face supplies 40 of them. Bars are coloured by trust tier.
- huggingface.co40
- api.llm-stats.com21