Share your thoughts, 1 month free Claude Pro on usSee more

Vision-Language Perception on VL Perception Benchmarks

85.5TextVQA

Base

Updated 1mo ago

Evaluation Results

Method	Links
Base 2026.01		85.5	86.5	82.9	76.7
Base 2026.01		85.5	86.4	82.9	77
FRISM 2026.01		85.5	86.5	83	77
DARE 2026.01		85.4	86.7	82.9	76.7
FRISM 2026.01		85.4	86.6	82.9	76.8
DARE 2026.01		85.4	86.4	82.9	76.9
TA 2026.01		85	86.5	82.8	76.8
IP-Merging 2026.01		84.8	86.1	82.6	77
TA 2026.01		84.4	86.6	82.5	76.6
TA 2026.01		84.1	86.4	82.3	76.4
Ties-Merging 2026.01		83.8	86.4	82.4	76.9
IP-Merging 2026.01		83.2	87.2	82.3	76.7
TA 2026.01		82.9	86.4	81.5	75.1
IP-Merging 2026.01		82.6	86	81.7	76.7
IP-Merging 2026.01		82.5	86.8	82	76.8
Base 2026.01		82.2	90.3	83.2	77.2
FRISM 2026.01		82.2	90.3	83.3	77.4
IP-Merging 2026.01		82.1	90.2	83.2	77.2
DARE 2026.01		82	90.7	83.4	77.4
TA 2026.01		81.8	90	83.1	77.6
Ties-Merging 2026.01		80.3	88.7	82.1	77.2
Ties-Merging 2026.01		74.3	84.3	74.7	65.4
TA 2026.01		68.6	72.6	67.7	61.9
TA 2026.01		62.8	62.6	65.4	70.7
FRISM 2026.01		61.6	86.3	74	-
IP-Merging 2026.01		61.3	86.2	73.8	-
Base 2026.01		61.2	87.1	74.2	-
TA 2026.01		61.2	86.7	74	-