WorldVQA: Results as Published

WorldVQA tests atomic, vision-centric world knowledge in multimodal large language models: whether a model can recognise and name what is shown in an image, with recall separated from reasoning. The dataset has 3,500 VQA pairs across 9 categories, from common head-class objects to long-tail rarities. A judge model (an open-weights OpenAI model by default) grades the answers, and the leaderboard reports the first eight categories because the People category is excluded. The page refers to a 50% accuracy threshold and does not name the metric explicitly.

Every number here is read from the same catalog file as the benchmark table, data version 2026103001, and is the number the table shows for that model and test. Where one page printed a model under several setups, each value is listed. The catalog is public in armanobosyan/sugra-llm-benchmarks; a wrong number can be reported from the table, with Report an error under the value, or as an issue in that repository.