Thyme-7B: benchmark results
Provider: Other.
Aggregate Rating 1583 ± 32, rank #1208 of 10916 rated models, from 44 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| DailyClue - Location Identification | 69 | Accuracy (%) on location identification (country and first-l | 100 |
| DailyClue | 46.25 | Overall accuracy (%) on DailyClue, 666 question-image pairs | 81.8 |
| AdaptMMBench - Real-world | 58 | Accuracy (%) on the Real-world domain of AdaptMMBench, enhan | 80 |
| VTC-Bench (Code Tools) - Measurement | 40 | Pass rate (%) on measurement (105 questions on size, positio | 75 |
| VTC-Bench (Code Tools) - Counting | 38.82 | Pass rate (%) on counting (85 questions), answering VTC-Benc | 66.7 |
| VTC-Bench (Interface Tools) - Chart | 33 | Pass rate (%) on chart reasoning (100 questions), answering | 61.1 |
| DiCoBench - Instance Commonality | 66.4 | Accuracy (%) on the instance-commonality task (re-identify o | 55.9 |
| AdaptMMBench - Key Step Coverage | 77.14 | Key step coverage (%): how far the adaptive-mode reasoning c | 50 |
| VTC-Bench (Code Tools) - Chart | 35 | Pass rate (%) on chart reasoning (100 questions), answering | 47.2 |
| DiCoBench - Category Commonality | 53.6 | Accuracy (%) on the category-commonality task (match objects | 44.1 |
| AdaptMMBench (Text Reasoning) | 51.92 | Overall accuracy (%) on the 1,300 AdaptMMBench samples (auxi | 40 |
| AdaptMMBench - Math | 50 | Accuracy (%) on the Math (without auxiliary-line problems) d | 35 |
Interactive version: theaggregate.ai/model?slug=thyme-7b · How It Works · Data refreshed daily, snapshot 2026-11-08.