BLOOMZ-7b1: benchmark results
BigScience's BLOOMZ 7.1B, BLOOM fine-tuned on the crosslingual xP3 task mixture for zero-shot instruction following across dozens of languages. Provider: BigScience. Released 2022-09-27. Access: Open.
Aggregate Rating 1150 ± 26, rank #1604 of 1640 rated models, from 18 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 45.2 | MC2 (%) (0-shot) | 30.1 |
| Hallucinations Leaderboard | 38.75 | Average Task Score (score) | 28.9 |
| Open LLM Leaderboard v1 - MMLU | 37.85 | Accuracy (%) (5-shot) | 19.6 |
| Open LLM Leaderboard v1 - ARC Challenge | 42.49 | Normalized accuracy (%) (25-shot) | 18.5 |
| Open LLM Leaderboard v1 - WinoGrande | 64.64 | Accuracy (%) (5-shot) | 17.6 |
| Open LLM Leaderboard v1 - HellaSwag | 63.01 | Normalized accuracy (%) (10-shot) | 16.3 |
| Greek MMLU - STEM | 34.27 | Accuracy (%) | 9.9 |
| Open LLM Leaderboard v1 - GSM8K | 0.08 | Accuracy (%) (5-shot) | 9.5 |
| Greek MMLU - Other | 32.3 | Accuracy (%) | 8.6 |
| Greek MMLU | 32.4 | Accuracy (%) | 4.9 |
| Greek MMLU - Humanities | 32.63 | Accuracy (%) | 2.5 |
| Greek MMLU - Social Sciences | 30.03 | Accuracy (%) | 2.5 |
Interactive version: theaggregate.ai/model?slug=bloomz-7b1 · How It Works · Data refreshed daily, snapshot 2026-11-07.