Cohere medium v20220720 (6.1B): benchmark results
Provider: Other.
Aggregate Rating 1105 ± 34, rank #10662 of 10916 rated models, from 32 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| RealToxicityPrompts | 0.01 | Toxic fraction | 97.6 |
| HELM Classic - LSAT | 21.16 | Exact Match (%) | 59.6 |
| HELM Classic - BLiMP | 82.3 | Exact Match (%) | 54.8 |
| HELM Classic - IMDB | 93.5 | Exact Match (%) | 50 |
| HELM Classic - WikiFact | 25.41 | Exact Match (%) | 48.5 |
| HELM Classic - MS MARCO TREC | 37.4 | NDCG@10 (%) | 40 |
| HELM Classic - Dyck | 51.07 | Exact Match (%) | 36.8 |
| HELM Classic - LegalSupport | 50.72 | Exact Match (%) | 33.8 |
| HELM Classic - Entity Data Imputation | 72.15 | Exact Match (%) | 33.3 |
| HELM Classic - MMLU | 27.88 | Exact Match (%) | 33.3 |
| HELM Classic - HellaSwag | 70.6 | Exact Match (%) | 32.3 |
| HELM Classic - BBQ | 34.37 | Exact Match (%) | 31.7 |
Interactive version: theaggregate.ai/model?slug=cohere-medium-v20220720-6-1b · How It Works · Data refreshed daily, snapshot 2026-11-08.