Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Multimodal Visual Question Answering on MMBench
Loading...
88.6
Score
Qwen2.5-VL
70.92
75.51
80.1
84.69
May 29, 2025
Jul 23, 2025
Sep 17, 2025
Nov 11, 2025
Jan 6, 2026
Mar 2, 2026
Apr 27, 2026
Score
Updated 2d ago
Evaluation Results
Method
Method
Links
Score
Qwen2.5-VL
Model Size=72B
2026.04
88.6
GIFT
backbone=Qwen3-VL 8B
2025.10
87.2
Greedy
backbone=Qwen3-VL 8B
2025.10
86.9
Greedy
backbone=Qwen2-VL 7B
2025.10
84.6
GIFT
backbone=Qwen2-VL 7B
2025.10
84.6
Qwen2.5-VL
Model Size=7B
2026.04
83.5
Qwen2.5-VL
Model Size=7B, Reprodu...
2026.04
82.47
ForeSight
Model Size=7B
2026.04
81.5
GIFT
backbone=LLaVA-1.5 13B
2025.10
75.8
Greedy
backbone=LLaVA-1.5 13B
2025.10
75.6
Greedy
backbone=LLaVA-1.5 7B
2025.10
73.1
GIFT
backbone=LLaVA-1.5 7B
2025.10
73.1
Qwen2-VL-2B-Instruct (DPO, HighAvg.)
% Train=33, Train set=...
2025.05
72.7
Qwen2-VL-2B-Instruct (DPO, LowAvg.)
% Train=33, Train set=...
2025.05
72.3
Qwen2-VL-2B-Instruct (Zeroshot)
% Train=0, Train set=Z...
2025.05
72
Qwen2-VL-2B-Instruct (DPO, HighVar.)
% Train=33, Train set=...
2025.05
72
Qwen2-VL-2B-Instruct (DPO, Random)
% Train=33, Train set=...
2025.05
71.9
Qwen2-VL-2B-Instruct (DPO, Full)
% Train=100, Train set...
2025.05
71.6
Feedback
Search any
task
Search any
task