BlogLLM Benchmark Scores as Published, Side by Side
Dense200: Results as Published
Every result the catalog holds for Dense200: 4 models from 4 developers, read from 1 page. Each number is as printed by the company named beside it; nothing is ranked or rescored. Models are listed by developer, newest first.
Area Multimodal UnderstandingMaintainer Not stated on the page readIntroduced 2026Data version 2026103001, as of 2026-10-30
Dense200 tests visual grounding in dense natural scenes: the model has to localise many objects in one image, and the bbox variant scores the predicted bounding boxes. The Mistral Large 4 page prints it as a percentage.
Not checked against a maintainer's page.
Dense200 (bbox)
| Model | Value | Printed by | Setup as printed |
|---|---|---|---|
| GPT-6 AstraOpenAI, released 2026-09-03 | 41.5% | Mistral AI, Mistral Large 4 (preview)checked 2026-10-15 | - |
| DeepSeek-V4.1-FlashDeepSeek, released 2026-09-10 | 3.3% | Mistral AI, Mistral Large 4 (preview)checked 2026-10-15 | - |
| Kimi K3Moonshot AI, released 2026-07-16 | 28.9% | Mistral AI, Mistral Large 4 (preview)checked 2026-10-15 | - |
| Mistral Large 4 (preview)Mistral AI, released 2026-10-06 | 42% | Mistral AI, Mistral Large 4 (preview)checked 2026-10-15 | - |
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.