Illustration, AI-generated from the story text. Tap a panel to read the story through it.

Baidu published ERNIE-5.5-Flash-Base on November 5th, a model with about 280 billion total (the page says 286 billion) and 14 billion active, pretrained on 33.5 trillion tokens.

It is the at the end of with no mid-training or , which is unusual for a Chinese lab and useful to anyone who wants to fine-tune from a clean starting point. Weights are on Hugging Face and ModelScope in FP8 and BF16. Baidu says it competes with DeepSeek V4 Flash and Xiaomi's MiMo V2.5, from its own chart. It ships under Baidu's own license; blogs say it excludes the EU and UK and is free below $50 million in revenue, which we could not check against the text. A post-trained version is promised.

It was trained at up to 128,000 of context and designed for up to one million in and .
Baidu releases the raw base of a 280-billion-parameter ERNIE model
Baidu published ERNIE-5.5-Flash-Base on November 5th, a mixture-of-experts model with about 280 billion total parameters (the Hugging Face page says 286 billion) and 14 billion active, pretrained on 33.5 trillion tokens. It is the checkpoint at the end of pretraining with no mid-training or post-training, which is unusual for a Chinese lab and useful to anyone who wants to fine-tune from a clean starting point. Weights are on Hugging Face and ModelScope in and BF16.
It was trained at up to 128,000 tokens of context and designed for up to one million in fine-tuning and inference. Baidu says it competes with DeepSeek V4 Flash and Xiaomi's MiMo V2.5, from its own chart. It ships under Baidu's own license; blogs say it excludes the EU and UK and is free below $50 million in revenue, which we could not check against the text. A post-trained version is promised. SourcesAB
Sources
- AHugging Face: baidu/ERNIE-5.5-Flash-Basehuggingface.co
- BThe Decoder: Baidu releases ERNIE 5.5 Flash Basethe-decoder.com
This story ran in the Morning Brief of November 9th 2026.