Ripple-65B Preview beats open US models in coding benchmark
ripple-65b preview, a fully apache 2.0 open us model, achieved 80.52 on terminalbench 2.1 coding benchmark, surpassing other open us models. this impacts the open-source ai community.
ricursive intelligence developed ripple-65b preview, an open language model with 65 billion parameters, using a dense transformer architecture. it was pretrained on 12 trillion tokens and achieved a score of 80.52 on the terminalbench 2.1 coding benchmark, outperforming other open us models. ripple-65b preview also scored 79.1% on swe-bench pro v2.0 hard and 71.67% on mcp atlas. the model is available on hugging face under the apache 2.0 license and cost $32 million to develop and train.