Vision & Language Benchmark

VisualWebBench

VisualWebBench, benchmark for understanding web pages/screenshots (GUI grounding and navigation). Values reported as percentage. Source referenced on the LiquidAI/d1-2B model card.

Category

Vision & Language

Score type

percent

Scale

100%

higher is better

Models tested

1

What is VisualWebBench?

VisualWebBench is a benchmark for understanding images combined with text. The official description says: "VisualWebBench, benchmark for understanding web pages/screenshots (GUI grounding and navigation)". Below we describe in simple words what benchmarks of this type generally examine.

What does it test?

VisualWebBench measures understanding images combined with text. In detail, it checks:

  • answering questions about images (figures, photos, charts, memes)
  • reading text inside images and connecting it to the question
  • visual reasoning such as counting, locating and comparing objects
  • combining what is seen with general knowledge

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: VisualWebBench, benchmark for understanding web pages/screenshots (GUI grounding and navigation). Values reported as percentage. Source referenced on the LiquidAI/d1-2B model card.

Frequently asked questions

What is the VisualWebBench benchmark?

VisualWebBench is a benchmark for understanding images combined with text. The official description says: 'VisualWebBench, benchmark for understanding web pages/screenshots (GUI grounding and navigation)'. Below we describe in simple words what benchmarks of this type generally examine.

What does VisualWebBench test?

VisualWebBench mainly measures understanding images combined with text. Concretely, it checks: answering questions about images (figures, photos, charts, memes); reading text inside images and connecting it to the question; visual reasoning such as counting, locating and comparing objects.

How is the VisualWebBench 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 VisualWebBench data and scores?

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/visualwebbench/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.

API

Benchmark metadata (JSON)

This page's benchmark record from the Open LLM Radar API

 Download

All model scores (JSON download)

Full score table as JSON; paginate with limit/offset params

 Docs

Open LLM Radar API docs

Swagger UI to explore and reuse all radar API endpoints

Model scores 1 model tested

Reset
# Model Score Date Source Verified
1 d1-2B
LiquidAI
96.2%
— LiquidAI/d1-2B model card (performance table) Unverified

Best score first. Weights filter: open. The bar shows the relative score between the best and lowest tracked model on this benchmark — it is not a percentage of solved tasks. Showing 1 of 1 scored models.