Baidu Releases ERNIE 5.5 Flash Base Open-Source MoE Model
First merged summary.
Baidu has released ERNIE-5.5-Flash-Base, a 280 billion parameter mixture-of-experts foundation model with 14 billion active parameters, pre-trained on 33.5 trillion tokens.
Baidu released ERNIE-5.5-Flash-Base as an open-source foundation model. The mixture-of-experts architecture has 280 billion total parameters with 14 billion active per input and supports up to 1 million token context. It was pre-trained on 33.5 trillion tokens and is available in FP8 and BF16 formats on Hugging Face. Benchmarks show competitive performance with DeepSeek-V4-Flash-Base in reasoning, code, and long-context tasks. The base model is intended for further development and fine-tuning.
- 280 billion total parameters, 14 billion active per input
- Pre-trained on 33.5 trillion tokens
- Open-source foundation model on Hugging Face
- FP8 and BF16 weights available
- Up to 1 million token context length
- Competitive benchmarks vs DeepSeek-V4-Flash-Base
What each outlet adds
beyond the shared facts- The DecoderDetails on 1M token context, 33.5T tokens pre-trained, FP8/BF16 formats, Hugging Face availability, and benchmarks
- MarkTechPostNames it ERNIE 5.5 Flash Base 280B MoE model as base for fine-tuning
- Vendors
- Baidu
- Products
- ERNIE 5.5 Flash Base
- AI models
- ERNIE-5.5-Flash-Base
Coverage
oldest first · one row per article · UTC- · 3d ago
The DecoderFirst report
Baidu releases ERNIE 5.5 Flash Base with 280 billion parameters
Open foundation model with 280 billion parameters and 14 billion active per input, pre-trained on 33.5 trillion tokens.
- · 3d ago
MarkTechPost
Baidu Releases ERNIE 5.5 Flash Base Model
Baidu launches 280B MoE base model with 14B active parameters as part of ERNIE 5.5 series.