---
format: "aidr-story-markdown/v1"
id: "aa04bedae4a24dfb34f60e20a374c85a7a2dd1d24f7bc3e44a823aeb0e5ea308"
canonical_url: "https://aidr.today/aa04beda?lang=en"
title: "PrimeSim: A Generalist World Model Simulator from Large-Scale Cross-Embodiment Demonstrations"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-11-04T00:00:00.000Z"
category: "Research"
topics: ["agent"]
source_urls: ["https://huggingface.co/papers/2611.04870"]
summary: "PrimeSim is a multiview, action-conditioned world model simulator developed through progressive midtraining on 30,110 hours of robot demonstrations across 22 platforms and posttraining that converts a pretrained video generator into an autoregressive simulator."
---

# PrimeSim: A Generalist World Model Simulator from Large\-Scale Cross\-Embodiment Demonstrations

> [Open the canonical story](<https://aidr.today/aa04beda?lang=en>)

**Published:** 2026-11-04T00:00:00.000Z
**Category:** Research
**Topics:** agent

## Summary

PrimeSim is a multiview, action\-conditioned world model simulator developed through progressive midtraining on 30,110 hours of robot demonstrations across 22 platforms and posttraining that converts a pretrained video generator into an autoregressive simulator\.

## Sources

- [Story source](<https://huggingface.co/papers/2611.04870>)