A small, post-trained Qwen3-8B model from the experiments in the "Recursive Language Models" paper:
https://arxiv.org/abs/2512.24601.
The model was trained on trajectories produced using a fixed system prompt. It assumes the environment/scaffold from our RLM repo.
We recommend using vLLM with our inference code at
https://github.com/alexzhang13/rlm to use it out of the box.
