Ricursive Intelligence 发布 Ripple-65B Preview 开源模型
一个 65B 的开源模型在 Terminal-Bench 2.1 上跑赢 501B 参数的竞品，还能在 DGX Spark 上本地运行，性价比故事值得看看。
Ricursive Intelligence 发布了开源模型 Ripple-65B Preview，在 Terminal-Bench 2.1 基准上得 80.5%，超过 Reflection 的 Beam-501B 的 80.1%。Beam-501B 每次任务激活 501B 参数中的 23B，而 Ripple 使用全部 65B 参数。官方称该模型从零开始训练，使用了 12 万亿 token 和 3200 万美元算力预算，支持 100 万 token 上下文，采用 Apache 2.0 许可证。
Ok, *this is impressive*: Ricursive Intelligence just released Ripple-65B Preview, an open-weight AI model that it says outperforms much larger US models on a key agentic coding benchmark. And even runs locally on my DGX Spark.
Ripple scored 80.5% on Terminal-Bench 2.1, narrowly ahead of Reflection’s Beam-501B at 80.1%, according to the companies’ reported results!
The size comparison comes with a detail: Ripple uses all 65B parameters for each task, while Beam activates 23B of its 501B parameters.
Ricursive says Ripple was trained on 12 trillion tokens with a $32 million compute budget - from scratch. The model supports up to 1 million tokens of context and is available under Apache 2.0.
