Ernie 5.1: benchmark results
Baidu's ERNIE 5.1 (May 2026), a sparse Mixture-of-Experts model for Chinese and English text with a 128K-token context. Provider: Baidu. Released 2026-05-09. Access: API.
Aggregate Rating 1696 ± 18, rank #202 of 1656 rated models, from 75 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| SuperCLUE-Writing - Mainstream Genre Writing | 84.88 | Score | 100 |
| Chatbot Arena (Text - Medicine & Healthcare) | 1493 | Arena Score | 93 |
| Chatbot Arena (Text - Business, Management & Financial Ops) | 1475 | Arena Score | 92.6 |
| SuperCLUE-Writing - Content Creativity | 89.35 | Score | 92.3 |
| SuperCLUE-Writing - Content Quality | 89.59 | Score | 92.3 |
| SuperCLUE-Writing - Overall | 87.57 | Score | 92.3 |
| Chatbot Arena (Text - Mathematical) | 1489 | Arena Score | 91 |
| Chatbot Arena (Text - Software & IT Services) | 1508 | Arena Score | 90.9 |
| Chatbot Arena (Text - Math) | 1478 | Arena Score | 89.7 |
| Chatbot Arena (Text - English) | 1478 | Arena Score | 89.3 |
| Chatbot Arena (Text - Hard Prompts (English)) | 1492 | Arena Score | 88.8 |
| ReLE - Education - Gaokao | 61.9 | Accuracy (%) | 88.5 |
Interactive version: theaggregate.ai/model?slug=ernie-5-1 · How It Works · Data refreshed daily, snapshot 2026-10-20.