xglm-1.7B: benchmark results
Meta's XGLM 1.7B, a multilingual autoregressive language model trained on 500 billion sub-tokens across a balanced set of languages. Provider: Meta. Released 2021-12-29. Access: Open.
Aggregate Rating 1126 ± 13, rank #1622 of 1640 rated models, from 79 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MMLU-by-task - High School Statistics | 47.69 | Accuracy (%) | 86 |
| MMLU-by-task - High School Mathematics | 30 | Accuracy (%) | 76.1 |
| MMLU-by-task - Abstract Algebra | 30 | Accuracy (%) | 52.2 |
| MMLU-by-task - Professional Medicine | 45.22 | Accuracy (%) | 48.2 |
| MMLU-by-task - High School Chemistry | 31.03 | Accuracy (%) | 43.8 |
| MMLU-by-task - College Physics | 23.53 | Accuracy (%) | 41.6 |
| MMLU-by-task - High School Computer Science | 33 | Accuracy (%) | 33.8 |
| MMLU-by-task - Prehistory | 32.1 | Accuracy (%) | 33.4 |
| MMLU-by-task - Anatomy | 32.59 | Accuracy (%) | 30.4 |
| MMLU-by-task - US Foreign Policy | 33 | Accuracy (%) | 30.4 |
| MMLU-by-task - Econometrics | 25.44 | Accuracy (%) | 26.9 |
| MMLU-by-task - High School Biology | 28.39 | Accuracy (%) | 26.5 |
Interactive version: theaggregate.ai/model?slug=xglm-1-7b · How It Works · Data refreshed daily, snapshot 2026-11-07.