Yi 6B (Base): benchmark results
01.AI's Yi-6B base model from the bilingual Yi series, trained from scratch on a 3-trillion-token multilingual corpus for language understanding and reasoning. Provider: 01.AI. Released 2023-11-02. Access: Open.
Aggregate Rating 1303 ± 13, rank #1472 of 1656 rated models, from 99 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MERA Original Paper - MathLogicQA | 38.2 | Accuracy (%; 5-shot, log-likelihood) | 100 |
| MERA Original Paper - SimpleAr | 95.1 | Exact Match (%; 5-shot, greedy generation) | 100 |
| MERA Original Paper - ruMultiAr | 18.9 | Exact Match (%; 5-shot, greedy generation) | 94.4 |
| MERA Original Paper | 35.4 | Total Score (%; mean of 17 problem-solving and exam tasks, M | 83.3 |
| MERA Original Paper - ruMMLU | 48.7 | Accuracy (%; 5-shot, log-likelihood) | 83.3 |
| MERA Original Paper - ruModAr | 41.6 | Exact Match (%; greedy generation) | 83.3 |
| MERA Original Paper - ruOpenBookQA | 59 | Accuracy (%; 5-shot, log-likelihood) | 83.3 |
| MERA Original Paper - ruWorldTree | 54.1 | Accuracy (%; 5-shot, log-likelihood) | 77.8 |
| Open Chinese LLM - WinoGrande | 64.56 | Accuracy (%) | 67.2 |
| Open Chinese LLM - C-Eval Semantic | 74.77 | Accuracy (%) | 64.8 |
| C-Eval | 72 | Average (%) | 62.7 |
| Open Chinese LLM - CMMLU | 56.54 | Accuracy (%) | 61.8 |
Interactive version: theaggregate.ai/model?slug=yi-6b-base · How It Works · Data refreshed daily, snapshot 2026-10-20.