Yi 6B (Chat): benchmark results
01.AI's 6B chat 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 2023-11-23. Access: Open.
Aggregate Rating 1340 ± 11, rank #1408 of 1656 rated models, from 233 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OpenCompass Knowledge - Common Sense - Chinese (CompassBench 2404) | 69.4 | Score (%) | 80 |
| HELM NaturalQuestions (Open) | 74.78 | F1 (%) | 73.3 |
| JustEval - Depth | 4.39 | Score (1-5) | 73.3 |
| OpenCompass Knowledge - Social Science - Chinese (CompassBench 2404) | 58.6 | Score (%) | 73.3 |
| SALAD-Bench Attack | 22.56 | Safety Score (%) | 72.7 |
| OpenCompass Knowledge - Humanities - Chinese (CompassBench 2404) | 45.3 | Score (%) | 70 |
| OpenCompass LLM - Knowledge - Chinese (CompassBench 2404) | 53.2 | Score (%) | 70 |
| OpenCompass Language - Traditional Cultural Understanding - Chinese (CompassBench 2404) | 47.7 | Score (%) | 66.7 |
| Open Chinese LLM - C-Eval Semantic | 75.32 | Accuracy (%) | 65.7 |
| OpenCompass Knowledge - Natural Science - Chinese (CompassBench 2404) | 39.6 | Score (%) | 63.3 |
| OpenCompass LLM - Knowledge (CompassBench 2404) | 58.1 | Score (%) | 63.3 |
| Open Chinese LLM - CMMLU | 56.6 | Accuracy (%) | 62.1 |
Interactive version: theaggregate.ai/model?slug=yi-6b-chat · How It Works · Data refreshed daily, snapshot 2026-10-20.