@Baidu_Inc released ERNIE 5.5 Flash Base.

This is a dev post classified by AGK Intelligence as Other (a launch), kept by the Dev Radar because it carries real work, not commentary.

@Baidu_Inc released ERNIE 5.5 Flash Base. An MoE model with 280B parameters total and 14B active per token and 1M context window. This is the checkpoint before any post-training, i.e. it is a starting point for fine-tunes rather than something you use on a regular basis. The interesting part in this model is the attention. Most layers cache one entry per 128 tokens and only six keep full context, which is how they make a million token window affordable. On their own pipeline it beats DeepSeek V4 Flash Base, Hy3 preview Base and MiMo V2.5 Base across knowledge and code, with MMLU at 93.3 a

Posted by Javier Bits • priv/acc (5k followers) 1 h ago · 0 likes · 132 views · view the original post on X. Kept by the Dev Radar as Other.

More dev work like this

• 就在刚刚，百度文心终于出手了，把自家基座模型文心 5.5 Flash Base 开源。 — @GitHub_Daily
• 🚀Baidu just open-sourced ERNIE-5.5-Flash-Base — @Chinazhidx
• JUST IN: @VerneRobotics released V0.5, an open-weight model for industrial robots. — @IlirAliu_
• 读英文 PDF，最耗神的往往不是内容本身，而是术语和长句总把节奏打断。 — @bkdgiffug
• Nanjing Iron and Steel (NISCO) has 300+ AI models, 25 business agents, and 1,000+… — @Huawei
• 搭 3D 场景时缺几个小道具，桌上的杯子、墙角的椅子，自己建模太费时间，去素材站翻半天也不一定有合适的。 — @GitHub_Daily
• 2 minutes of continuous AI video with spoken dialogue, generated at home on two NVIDIA… — @majewskizby
• 油猴撑了好几年其实也够用，可真遇上页面关了脚本还得接着跑、定时任务又不想死盯标签页的活儿，它就有点接不住了。 — @bkdgiffug

Every post is read and classified by AGK Intelligence: what it is, which market it belongs to, and whether the link is a real tool. 31.7k posts from 5k X accounts over the last 21 days, 9.3k tools, 12 markets. New posts are collected several times an hour and the full ranking is recomputed every hour — last update 2026-11-05 10:13 UTC. Full method.