百度 ERNIE 5.5 Flash Preview 上架 LMArena，权重 11 月 25 日开源
百度把 ERNIE 5.5 Flash 丢进 LMArena 让大家盲测了，280B 参数只激活 14B，速度还超 200 tokens/sec。写代码、跑长文档的可以上 Battle Mode 试试水，11 月 25 日开源权重后自己部署也很划算。
百度 ERNIE 5.5 Flash Preview 现已进入 LMArena，用户可在 Battle Mode 中用编程、Web 开发等真实任务对比测试，Code Arena 分数随后公布。模型总参数 280B、每 token 激活 14B，采用面向高效推理的新架构，支持 1M-token 上下文窗口，生成速度超过 200 tokens/sec。
官方称其在 Terminal-Bench、AutomationBench 等基准上保持有竞争力的成绩，同时解码速度不减，应用场景覆盖编程、3D 建模、知识工作和内容创作。模型权重将于 11 月 25 日前发布，是 ERNIE 家族最新的开源权重基础模型。
200+ tokens/sec 的解码速度配合 1M 上下文，使它在长文档处理和高并发场景中具备部署吸引力；MoE 式的 14B 激活参数设计也意味着推理成本显著低于稠密 280B 模型，对自建服务的企业开发者值得关注。
ERNIE 5.5 Flash Preview by @Baidu_Inc is now in the Arena. Head over to test it in Battle Mode with your toughest coding, web development, and real-world tasks to see how it does. Scores for Code Arena coming soon. Baidu Inc. @Baidu_Inc ERNIE 5.5 Flash Preview just landed! 🔥 Built on a new architecture designed for efficient inference, the latest open-weight foundation model in the ERNIE family delivers stronger performance in coding, 3D modeling, knowledge work and content creation. It keeps decoding fast while posting competitive scores on Terminal-Bench, AutomationBench, and other benchmarks. Here's the rundown: - 280B total parameters, 14B active per token - 1M-token context window - Generation speed above 200 tokens/sec Model weights will be released by November 25. 🔗 View Quoted Tweet📊 9 ⚡
This report is categorized under AITOP's "Models" desk, focusing on China's AI industry and domestic foundation model ecosystem. Within our cross-regional analytical framework, it directly maps to US-China foundation model benchmarks and commercialization comparison and AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include ERNIE, Baidu, 开源模型, LMArena.
Source attribution: curated from lmarena.ai (@lmarena_ai).
