Today for AI
HOT RADAR
llmHEAT 11.2°

AI CURATED HOT STORY · 11/4/2026

China Merchants Releases Lion-ER-35B-A3B Embodied Reasoning Model with Capability-Driven Data Engine

1 reports archived1 independent sourcesupdated 11/4/2026, 18:08:38
Synthesis & Latest Updates
1 Sources Cross-Validated

The China Merchants Lion Mountain Lab released the open-source embodied reasoning VLM 'Lion-ER-35B-A3B'. Utilizing a MoE architecture (35B total, 3B active), it shifts from data accumulation to a 'capability-driven' data engine, decomposing tasks into 8 domains and 39 fine-grained capabilities for iterative optimization. The model achieves improvements across ten mainstream embodied benchmarks, surpassing larger models in key areas by enhancing robots' understanding of 3D space, object states, and task progress.

LATEST/The China Merchants Lion Mountain Lab released the open-source embodied reasoning VLM 'Lion-ER-35B-A3B'. Utilizing a MoE architecture (35B total, 3B active), it shifts from data accumulation to a 'capability-driven' data engine, decomposing tasks into 8 domains and 39 fine-grained capabilities for iterative optimization. The model achieves improvements across ten mainstream embodied benchmarks, surpassing larger models in key areas by enhancing robots' understanding of 3D space, object states, and task progress.

Hourly heat curve

4 fully observed hours
Now
11.2
Peak
11.819:00
24h change
–

The line compares only sources observed throughout; gap hours are interpolated to keep the trend continuous.

Coverage timeline

Total 1 reports · Latest first
  1. 量子位官网T2·78 pts

    China Merchants Releases Lion-ER-35B-A3B Embodied Reasoning Model with Capability-Driven Data Engine

    • Uses a MoE architecture with 35B total parameters but only 3B active per token, balancing performance and computational cost.
    • Key innovation is the 'capability-driven embodied data engine', decomposing reasoning into 8 primary domains and 39 secondary capabilities for positive design rather than blind data collection.
    • Achieved improvements across all metrics in ten mainstream embodied benchmarks, surpassing larger models in several core capabilities.
    • Model is open-sourced on Hugging Face and supports major inference frameworks like vLLM and SGLang for easy integration.