multimodal benchmark
VLMsAreBiased
VLMsAreBiased evaluates whether vision-language models rely on visual evidence or fall back on language priors when answering.
Updated Aug 7, 2026
Published models2
Registry coverage2
MetricScore
EvidenceB
VLMsAreBiased leaderboard
Sorted by the source-provided rank. Higher score is better according to the registry.
| 1 | BY | 83.6% | 100.0% | 2 | C | |
| 2 | BY | 68.3% | 0.0% | 2 | C |
Score distribution
Top published rows on the benchmark's original scale.
VLMsAreBiased highlights
The top published results on this benchmark's own scale.
What is VLMsAreBiased?
Definition and scoring fields from the benchmark registry.
VLMsAreBiased evaluates whether vision-language models rely on visual evidence or fall back on language priors when answering.
Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level B.
- Family
- VLMsAreBiased
- Modality
- multimodal
- Primary category
- multimodal
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- vlmsarebiased|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
FAQ
Common questions about VLMsAreBiased.
Which model scores highest on VLMsAreBiased?
Seed 2.1 Pro is currently ranked first with 83.6%.
What does VLMsAreBiased measure?
VLMsAreBiased evaluates whether vision-language models rely on visual evidence or fall back on language priors when answering.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
2 unique published model results are currently shown.
Does this benchmark affect the overall score?
This benchmark is preserved as source-native evidence but is not eligible for the current overall score.