Yi Lightning: benchmark results
01.AI's fast proprietary MoE flagship, API-only, which debuted in the Chatbot Arena top 10 at launch (October 2024). Provider: 01.AI. Released 2024-10-16. Access: API.
Unified ELO 1669 ± 19, rank #402 of 2656 rated models, from 152 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OpenCompass LLM - Language (CompassBench 2411) | 63.7 | Score (%) | 100 |
| OpenCompass LLM - Language - Chinese (CompassBench 2411) | 66.1 | Score (%) | 100 |
| OpenCompass LLM - Language - English (CompassBench 2411) | 61.2 | Score (%) | 100 |
| OpenCompass Reasoning - Common Sense - Chinese (CompassBench 2411) | 71.2 | Score (%) | 100 |
| OpenCompass Reasoning - Social - Chinese (CompassBench 2411) | 68.6 | Score (%) | 100 |
| OpenCompass Language - Dialogue - Chinese (CompassBench 2411) | 70.1 | Score (%) | 98.5 |
| Ru Arena Hard | 93.46 | Win Rate (%) | 98.5 |
| OpenCompass Language - Dialogue - English (CompassBench 2411) | 67.3 | Score (%) | 97 |
| OpenCompass Language - NLP - Chinese (CompassBench 2411) | 65.4 | Score (%) | 97 |
| OpenCompass Reasoning - Common Sense - English (CompassBench 2411) | 67.8 | Score (%) | 97 |
| OpenCompass Reasoning - Humanities - English (CompassBench 2411) | 67.8 | Score (%) | 97 |
| OpenCompass Reasoning - Science and Engineering - Chinese (CompassBench 2411) | 68 | Score (%) | 97 |
Interactive version: theaggregate.ai/model?slug=yi-lightning · How It Works · Data refreshed daily, snapshot 2026-09-19.