BLINK
BLINK - multimodal benchmark for visual perception tasks.
Category
Multimodal
Score type
percent
Scale
100%
higher is better
Models tested
5
What is BLINK?
BLINK is a benchmark for handling several input types together (images, documents, text). The official description says: "BLINK - multimodal benchmark for visual perception tasks". Below we describe in simple words what benchmarks of this type generally examine.
What does it test?
BLINK measures handling several input types together (images, documents, text). In detail, it checks:
- questions that mix text with images or document pages
- reading and extracting information from screenshots and scans
- understanding layout, titles, logos and figures in real documents
- staying accurate when the text content alone is not enough
How to read the score
The score is accuracy: the percentage of tasks answered or solved correctly. 100% means every task was solved perfectly; higher is better.
Official methodology note: BLINK - multimodal benchmark for visual perception tasks.
Frequently asked questions
What is the BLINK benchmark?
BLINK is a benchmark for handling several input types together (images, documents, text). The official description says: 'BLINK - multimodal benchmark for visual perception tasks'. Below we describe in simple words what benchmarks of this type generally examine.
What does BLINK test?
BLINK mainly measures handling several input types together (images, documents, text). Concretely, it checks: questions that mix text with images or document pages; reading and extracting information from screenshots and scans; understanding layout, titles, logos and figures in real documents.
How is the BLINK score calculated?
The score is accuracy: the percentage of tasks answered or solved correctly. 100% means every task was solved perfectly; higher is better. The declared maximum score of this benchmark is 100.0.
Where do I get BLINK data and scores?
The official source is https://arxiv.org/abs/2404.12390. The official benchmark page/reference is linked on this page. Every score on Open LLM Radar is also available as open data: download the scores JSON (http://open-llm-radar-api:8000/api/v1/benchmarks/blink/scores) or pull it directly from the Open LLM Radar API with any HTTP client.
Benchmark data & downloads
Where to get, inspect and reuse the data behind this benchmark.
Official benchmark page / paper
The benchmark maintainer's website, paper or leaderboard
Benchmark metadata (JSON)
This page's benchmark record from the Open LLM Radar API
All model scores (JSON download)
Full score table as JSON; paginate with limit/offset params
Open LLM Radar API docs
Swagger UI to explore and reuse all radar API endpoints
Model scores 5 models tested
| # | Model | Score | Date | Source | Verified |
|---|---|---|---|---|---|
| 1 |
LFM3-VL-2B
LiquidAI
|
71.1%
|
— | Report | Verified |
| 2 |
d1-2B
LiquidAI
|
66.2%
|
— | LiquidAI/d1-2B model card (performance table) | Unverified |
| 3 |
Gemma 3 27B PT
Google DeepMind
|
39.6%
|
27.02.2025 | Report | Verified |
| 4 |
Gemma 3 4B PT
Google DeepMind
|
38%
|
27.02.2025 | Report | Verified |
| 5 |
Gemma 3 12B PT
Google DeepMind
|
35.9%
|
27.02.2025 | Report | Verified |