Skip to content
EVENT / TOPICActive

Baidu Open-Sources ERNIE 5.5 Flash Base, a 280B-Parameter MoE Foundation Model

2 reports2 sources11/5/2026 updated

Current overview (not historical)

AI overview

Baidu has open-sourced ERNIE 5.5 Flash Base on Hugging Face, a 280-billion-parameter mixture-of-experts foundation model that activates 14 billion parameters per token. The model was pretrained on 33.5 trillion tokens with a maximum pretraining context of 128K tokens, while fine-tuning and inference support contexts up to 1 million tokens. Baidu said it achieves state-of-the-art results among same-sized models on multiple benchmarks. The Decoder reported that the release contains no fine-tuning or post-training, so developers must determine the model’s direction themselves; a post-training version is expected to follow.

Written with AI from the reporting · updated 11/5/2026

Latest developmentNov 6, 01:56
Baidu Releases ERNIE 5.5 Flash Base Open-Source Foundation Model with 280 Billion Parameters
Explore the timeline, relations and evidence
Expand the detailed fact route and sources

Story route

A station-by-station route from the first public evidence onward; every station carries its own evidence and uncertainty.

  1. 1Unclassified fact stationNov 5, 00:00recorded Nov 5, 17:54revised

    百度开源文心5.5 Flash Base基座模型

    Read original source ↗ QbitAI · AI Industry & ResearchReplay this moment in the explorer
    Evidence & revisions · 2 source records
    Subject：
    百度
    Action：
    开源
    Object：
    文心5.5 Flash Base（ERNIE 5.5 Flash Base）
    Occurred (fact side)：
    Nov 5, 00:00
    QbitAI · AI Industry & Research ↗Report pageRevision 2 · currently revised
    Published (valid side)：
    Nov 5, 15:15
    First seen (known side)：
    Nov 5, 17:54
    The Decoder ↗Report pageRevision 2 · currently revised
    Published (valid side)：
    Nov 6, 01:56
    First seen (known side)：
    Nov 6, 02:13
    Fact stationRevised

    Copy may have been revised later; time replay is bounded. Layer labels group reading only, not causation between layers.

Uncertainty marks flag where to check the sources; they are not proven causation.

Report timeline

Explore every side through dated source reports.

Nov 6
  1. The DecoderSource report revised
    Baidu Releases ERNIE 5.5 Flash Base Open-Source Foundation Model with 280 Billion Parameters

    Report first recorded Nov 6, 02:13

    Baidu has released ERNIE 5.5 Flash Base, an open foundation model with a mixture-of-experts architecture totaling 280 billion parameters, 14 billion active per input. It was pre-trained on 33.5 trillion tokens with up to 128,000-token training context and supports up to one million tokens during fine-tuning and inference. Weights are available on Hugging Face in FP8 and BF16. The base model ships without fine-tuning or post-training, with a post-trained version expected later. Baidu says it competes with similarly sized open foundation models such as DeepSeek V4 Flash and MiMo V2.5 in reasoning, code, and long-context tasks.

    Read original source ↗
Nov 5
  1. QbitAI · AI Industry & ResearchSource report revised
    Original-language report — open source

    Report first recorded Nov 5, 17:54

    Read original source ↗

Event activity trend

Latest comparable
17
Comparable peak
17Nov 6, 02:00
24h change
–

The chart compares consistently observed participants only. Hover or tap to inspect each hour; use arrow keys to move between observations.

Event discussion

Reader opinions are hosted on GitHub Discussions; posting requires a GitHub account. Comments are not verified facts.

Open GitHub Discussions