ERNIE 5.0 Thinking Preview — benchmark results
Baidu's extended-thinking preview variant of the multimodal ERNIE 5.0 flagship. Provider: Baidu. Released 2026-01-01. Access: API.
Unified ELO 1694 ± 32, rank #282 of 1776 rated models, from 43 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AA LiveCodeBench | 81.16 | Pass@1 (%) | 93.3 |
| AA Global-MMLU-Lite - English | 92.75 | Accuracy (%) | 83.2 |
| AA MMLU-Pro | 83.03 | Accuracy (%) | 83.1 |
| AA AIME 2025 | 85 | Accuracy (%) | 82.8 |
| CritPt | 1.4 | Accuracy (self-reported) | 80.1 |
| AA CritPt | 1.43 | Accuracy (%) | 75.1 |
| AA TAU-2 Bench | 83.92 | Accuracy (%) | 74.8 |
| AA Global-MMLU-Lite | 86.5 | Accuracy (%) | 73.1 |
| AA Humanity's Last Exam | 12.65 | Accuracy (%) | 69.4 |
| AA GPQA Diamond | 77.68 | Accuracy (%) | 68.4 |
| AA Omniscience - Software Engineering (SWE) - R | 20 | Accuracy (%) | 66.6 |
| AA Terminal-Bench Hard | 25 | Accuracy (%) | 66 |
Interactive version: theaggregate.ai/model?slug=ernie-5-0-thinking-preview · How the rankings work · Data refreshed daily, snapshot 2026-07-22.