Ripple-65B Preview: U.S. Open-Weight SOTA on Agentic Coding

Recursive Intelligence released Ripple-65B Preview, an open-weight language model that achieves state-of-the-art performance on agentic coding tasks, scoring 80.52 on TerminalBench 2.1 and outperforming significantly larger competitors. The model was trained from scratch for $32 million—one-tenth the estimated cost of previous open-weight SOTA models—using efficient training methods including custom synthetic data pipelines and optimized reinforcement learning techniques. Ripple-65B Preview supports up to 1M context length and is now available on multiple platforms including Hugging Face and Together AI.