@StanfordAILab’s @talaei_shayan shipping incredible work! 🚀

TL;DR · AI 摘要
斯坦福AI实验室发布Ripple-65B模型,在agentic coding任务中超越多个更大模型。
核心要点
- Ripple-65B在Terminal-Bench 2.1测试中超越Muse Spark Open-816B等更大模型
- 仅由5人团队训练完成65B参数模型
- 该模型是美国首个在agentic coding任务中达到SOTA的开源权重模型
结构提纲
按章节快速跳转。
- §模型发布
宣布Ripple-65B Preview的正式发布及其性能突破
- ·技术指标
65B参数规模及在Terminal-Bench 2.1测试中的SOTA表现
- ›训练团队
由五人研究团队从零开始训练模型的创新方法
- ·性能对比
超越Muse Spark Open-816B等更大模型的基准测试结果
- ›应用场景
在agentic coding任务中展现显著优势
思维导图
用一张图看清主题之间的关系。
查看大纲文本(无障碍 / 无 JS 友好)
- Ripple-65B模型发布
- 技术亮点
- 65B参数规模
- 开源权重模型
- agentic coding SOTA
- 训练信息
- 五人团队训练
- 从零开始预训练
- 性能对比
- 超越816B参数模型
- Terminal-Bench 2.1基准
金句 / Highlights
值得收藏与分享的关键句。
Ripple sets a new state-of-the-art for U.S. open-weight models in agentic coding tasks
trained it from scratch with a core team of five researchers
surpassing significantly larger models including Muse Spark Open-816B
Stanford AI Lab on X: "@StanfordAILab’s @talaei_shayan shipping incredible work! 🚀" / X
@StanfordAILab
’s
@talaei_shayan
shipping incredible work! 🚀
Shayan Talaei
11h
Really happy to finally share Ripple-65B Preview! We trained it from scratch with a core team of five researchers, and it sets a new U.S. open-weight SOTA on Terminal-Bench 2.1 for agentic coding. Three choices that shaped how we built it, and a few things I learned along the
Quote
Ricursive Intelligence
@RicursiveAI
12h
Today we release Ripple-65B Preview, our 65B dense LLM we pretrained from scratch. Ripple sets a new state-of-the-art for U.S. open-weight models in agentic coding tasks, surpassing significantly larger models including Muse Spark Open-816B, Beam-501B, Inkling-975B, and Nemotron
9:10 PM · Nov 4, 2026
·
5,203
Views
4
16