# Baidu's ERNIE 5.1 AI model cost 94% less to train than comparable systems.

*genai · news · 2026-05-11 · Decrypt*

## Key points

- ERNIE 5.1 cost 94% less to train than comparable large-scale AI models.
- Baidu used 'multi-dimensional elastic pre-training' to shrink ERNIE 5.1's parameters while retaining performance.
- On LMArena, ERNIE 5.1 ranked fourth globally and first among Chinese models for search tasks.
- Baidu implemented a four-stage MOPD pipeline to unify specialist models without skills trade-off.
- ERNIE 5.1 nearly matches leading western models on GPQA and scores 99.6% on AIME26 with tools.

**Companies:** Baidu
**Countries:** China

[Read the full story on Decrypt](https://decrypt.co/367493/baidu-ai-model-cost-less-train-beating-everyone-china)

---

Canonical: https://newsio.io/n/cdf2d362-e015-4f64-a35a-3faa100e48d7/baidus-ernie-5-1-ai-model-cost-94-less-to-train-than-comparable-systems-the-tric
Summarized by Newsio from Decrypt. https://newsio.io/how-it-works