We open-source 200+ model checkpoints, including a multi-task Newt agent trained on 200 tasks simultaneously. We are excited to see what the community will do with these models, and hope that our release will encourage other research labs to open-source their checkpoints as well. This section aims to provide further details about the released models.
As one of the first major releases of model checkpoints for reinforcement learning, use of our TD-MPC2 and Newt checkpoints is fairly open-ended. We envision that our checkpoints will be useful for researchers interested in training, finetuning, evaluating, and analyzing single-task and multitask models on any of the 200 continuous control tasks that we release models for. However, we also expect the community to discover new use cases for these checkpoints.
Direct use
Model checkpoints can be loaded using the official implementation, and then be used to reproduce our results and/or generate trajectories for any of the supported tasks.
Out-of-scope use
We do not expect our model checkpoints to reliably generalize to new (unseen) tasks as is. Such model usage will most likely require some amount of fine-tuning on target task data.
If you find our work useful, please consider citing the paper as follows:
BibTeX:
@misc{Hansen2025Newt,
title={Learning Massively Multitask World Models for Continuous Control},
author={Nicklas Hansen and Hao Su and Xiaolong Wang},
booktitle={Preprint},
url={https://www.nicklashansen.com/NewtWM},
year={2025}
}
