DBRX Instruct: benchmark results
Provider: Databricks. Released 2024-03-27.
Aggregate Rating 1426 ± 26, rank #4858 of 10916 rated models, from 462 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| HELM AIR-Bench 2024 - #6.24: Accounting | 63.33 | Refusal Rate (%) | 84.9 |
| HELM MMLU - Marketing | 94.02 | EM | 74.4 |
| HELM MMLU - Human Sexuality | 87.79 | EM | 73.1 |
| HELM MMLU - Computer Security | 83 | EM | 70.5 |
| HELM AIR-Bench 2024 - #39-40.33: Characterization of identity - Mental Characteristics | 96.67 | Refusal Rate (%) | 69.8 |
| HELM AIR-Bench 2024 - #6: Advice in Heavily Regulated Industries | 15.33 | Refusal Rate (%) | 69.8 |
| HELM Lite - MedQA | 69.38 | Exact Match (%) | 67.8 |
| HELM AIR-Bench 2024 - #6.23: Medical/Pharmaceutical | 10 | Refusal Rate (%) | 65.7 |
| HELM MMLU - Miscellaneous | 91.06 | EM | 64.1 |
| HELM AIR-Bench 2024 - #17.2: Suggestive | 63.33 | Refusal Rate (%) | 64 |
| HELM MMLU - Us Foreign Policy | 93 | EM | 62.8 |
| HELM MMLU - World Religions | 87.13 | EM | 62.2 |
Interactive version: theaggregate.ai/model?slug=dbrx-instruct · How It Works · Data refreshed daily, snapshot 2026-11-08.