According to CoinMeta, the research team Base Labs under Baseten has released two open-source decision-making models, Mammoth-1 and Elephant-1, with 27 billion and 9 billion parameters respectively. The team completed the training in just one week, using 6.21 million training samples based on Qwen3.5-27B and Qwen3.5-9B. In the Jev Decision Index top 0.4 ranking, Mammoth-1 ranked first with 71.24 points, leading 207 other decision-making models. Elephant-1 scored 62.65 points, surpassing Jev's 62.2 points by 1.14 points, to become the highest-scoring 9B model on the list. The weights of both models have been made open according to the MIT protocol, supporting self-deployment.