Dolphin-v2: benchmark results
Provider: Cognitive Computations.
Aggregate Rating 1702 ± 51, rank #321 of 10552 rated models, from 32 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MPDocBench-Parse - Heading Hierarchy | 50.04 | Tree-edit-distance similarity (0-100; predicted Markdown hea | 94.4 |
| MPDocBench-Parse - Table Recognition | 83.84 | TEDS (0-100; recognized HTML tables) | 83.3 |
| MPDocBench-Parse - Truncated Table Merging | 64.28 | TEDS (0-100; tables split across pages, matched to the merge | 83.3 |
| OmniDocBench 1.5 | 88.71 | Overall (mixed sources) | 55.7 |
| OmniDocBench v1.6 Full - Reading Order Edit | 0.15 | Normalized edit distance | 45.7 |
| MPDocBench-Parse | 73.14 | Overall score (0-100; mean of the eight task scores, with th | 44.4 |
| OmniDocBench v1.6 Full - Table TEDS | 84.4 | Table TEDS (%) | 42.9 |
| OmniDocBench v1.6 Full - Formula CDM | 91.01 | Formula CDM (%) | 40 |
| PureDocBench - Clean - Formula CDM | 59.8 | Formula Character Detection Matching score (0-100) between t | 36.8 |
| PureDocBench - Clean - Reading Order Edit Distance | 42.9 | Reading-order normalized edit distance (%) between the predi | 36.8 |
| OmniDocBench v1.6 Full - Overall | 89.5 | Overall (%) | 34.3 |
| PureDocBench - Clean - Table TEDS | 72.12 | Table Tree-Edit-Distance-based Similarity (0-100) between th | 33.3 |
Interactive version: theaggregate.ai/model?slug=dolphin-v2 · How It Works · Data refreshed daily, snapshot 2026-11-16.