Baidu releases ERNIE 5.5 Flash Base with 280 billion parameters as an open-source foundation model
Baidu released ERNIE 5.5 Flash Base, a 280-billion-parameter open-source MoE model pre-trained on 33.5 trillion tokens, shipped without fine-tuning.
Key Takeaways
- Key Highlight：Baidu released ERNIE 5.5 Flash Base, a 280-billion-parameter open-source MoE model pre-trained on 33.5 trillion tokens, shipped without fine-tuning.
- Key Topics：Baidu, ERNIE, Flash, Base
Baidu has released ERNIE 5.5 Flash Base, an open-source mixture-of-experts foundation model with 280 billion parameters, pre-trained on 33.5 trillion tokens. The model is deliberately shipped without fine-tuning, leaving downstream task customization to developers.
The release positions Baidu in the open-weight LLM race, where large Chinese and US labs publish base models for community adaptation. By skipping instruction tuning, Baidu lets developers apply their own alignment and domain-specific training.
The move could broaden access to a large-scale Chinese-language model, but its practical value depends on how well the base model performs after community fine-tuning. It also adds pressure on other labs to match parameter counts and open licensing.
This page provides an editorial summary based on publicly available information. It is not a republished article. Use the source link below for the original report.