# Allen Institute Releases MolmoBot — Robots Trained Entirely in Simulation Now Work in the Real World Without Fine-Tuning

*2026-03-14 06:04 UTC · Technology · 1 min read*

> Ai2 releases MolmoBot, a robot manipulation model trained purely in simulation that works on real robots without any fine-tuning.

The Allen Institute for AI (Ai2) has released MolmoBot, an open robotic manipulation model trained entirely on synthetic simulation data that transfers directly to real-world robots with zero fine-tuning. The breakthrough challenges a long-held assumption in robotics that virtual training alone is not sufficient for real-world performance.

MolmoBot was trained on 1.8 million simulated robot trajectories spanning over 100,000 environments and 30,000 unique objects. In testing, the model successfully operated a Franka FR3 robotic arm and a Rainbow Robotics RB-Y1 mobile manipulator, performing manipulation tasks on unseen objects without any real-world data collection.

Ai2 has open-sourced the entire stack, including MolmoSpaces, a large-scale simulation ecosystem for embodied AI. The release could dramatically reduce the cost and time required to train capable robots, eliminating the expensive real-world data collection bottleneck that has slowed the robotics industry.

## Sources

- [allenai.org](https://allenai.org/blog/molmobot)
- [roboticsandautomationnews.com](https://roboticsandautomationnews.com/2026/03/12/ai2-claims-breakthrough-in-training-robots-entirely-in-simulation/99578/)

---
TechTools News · https://techtools.cz/tools/news/article/785-allen-institute-releases-molmobot-robots-trained-entirely-in-simulation-now
Česká verze: https://techtools.cz/tools/news/clanek/806-allen-institute-vydal-molmobot-roboti-trenovani-ciste-v-simulaci-funguji-v
