---
title: "LionRock AI Lab open-sources 35B embodied reasoning model for robotics"
slug: "lionrock-ai-lab-open-sources-35b-embodied-reasoning-model-for-robotics"
date: "2026-11-05"
category: "AI & Automation"
author: "Dr. Kai Wei"
correspondent: "Dr. Kai Wei"
correspondentRole: "Lead Frontier AI Correspondent"
readingTimeMinutes: 2
tags: ["AI","News"]
originalSourceUrl: "https://pandaily.com/china-merchants-lionrock-ai-lab-lion-er-35b-a3b-embodied-reasoning-model-open-source"
canonicalUrl: "https://aurav.ae/lionrock-ai-lab-open-sources-35b-embodied-reasoning-model-for-robotics"
publisher: "Aura-V Intelligence (aurav.ae)"
---

# LionRock AI Lab open-sources 35B embodied reasoning model for robotics

> China Merchants' LionRock AI Lab released the Lion-ER-35B-A3B model, a 35-billion-parameter vision-language system for robot reasoning.

## Key Takeaways (TL;DR)

- LionRock AI Lab open-sourced Lion-ER-35B-A3B under Apache 2.0 on Hugging Face.
- The mixture-of-experts model has 35 B parameters, 3 B active per token, and outperforms its Qwen3.6-35B-A3B base on ten embodied benchmarks.
- Training used a capability-driven data engine with 312,380 curated examples and a feedback loop from a 4.3 M record pool.

China Merchants' LionRock AI Lab has released the Lion-ER-35B-A3B model, an embodied reasoning vision-language system aimed at helping robots interpret 3D space, object states and task progress, according to a report on [Pandaily](https://pandaily.com/china-merchants-lionrock-ai-lab-lion-er-35b-a3b-embodied-reasoning-model-open-source). The weights are hosted on Hugging Face under an Apache 2.0 license.

Lion-ER (LionRock Embodied Reasoner) is a mixture-of-experts architecture with roughly 35 billion total parameters and about 3 billion active per token, fine-tuned from Qwen3.6-35B-A3B. The lab says it improves on its base across ten public embodied and visual reasoning benchmarks, achieving an equal-weight average score of 65.29. Notable results include a Where2Place spatial placement score of 64.28 (up 9.30 points) and an ERQA score of 64.38 (up 6.19 points).

The team built a capability-driven embodied data engine rather than scaling raw data. They mapped reasoning into eight top-level domains and 39 finer skills, generating 312,380 supervised examples weighted by capability need. An evaluation-driven loop then feeds failures back into data production from a pool of about 4.3 million records across 31 data sources.

The model card notes that Lion-ER is intended for research, has not been validated for safety-critical decisions or autonomous robot control, and that benchmark gains may not transfer to new environments. The same methodology has been applied to smaller 4 B and 9 B backbones.

Our read: The open-source release lowers the barrier for Chinese developers to experiment with high-capacity embodied reasoning, but the modest benchmark gains and explicit safety caveats suggest it is a research-grade tool rather than a production-ready robot controller.

---

*Original Think-Tank Source: [China Merchants' LionRock AI Lab Open-Sources Lion-ER-35B-A3B Embodied Reasoning Model](https://pandaily.com/china-merchants-lionrock-ai-lab-lion-er-35b-a3b-embodied-reasoning-model-open-source)*