Ricursive Intelligence Launches Ripple-65B Preview Model

• •Ripple-65B Preview scores 80.52 on TerminalBench 2.1, surpassing larger models.
• •Developed for $32 million, significantly less than competitors' costs.
• •Available on multiple platforms including Hugging Face and GMI Cloud.

Ricursive Intelligence has unveiled Ripple-65B Preview, a 65 billion parameter dense Transformer model pretrained on 12 trillion tokens. The model achieves a score of 80.52 on TerminalBench 2.1, outperforming larger models like Reflection Beam 501B and NVIDIA Nemotron 3 Ultra 550B. Ripple-65B Preview is designed for efficient training and serving on NVIDIA hardware, with a focus on reducing infrastructure complexity. The model's development cost was significantly lower than competitors, at $32 million compared to estimates of up to $1 billion for similar models. The release is available on platforms such as Hugging Face and GMI Cloud, with a focus on agentic coding tasks. The architecture allows for low-latency token generation and strong reasoning performance. This release marks a significant advancement in AI model efficiency and performance, particularly for local device deployment.