Robot AI model
WALL-OSS
WALL-OSS couples a Qwen2.5-VL-3B backbone to action generation and multimodal reasoning. Wall-X code uses Apache-2.0, but separately stated checkpoint license terms were not established.
Records last verified Oct 13, 2026 · Send a correction
- Organization
- X Square Robot
- Type
- VLA
- Released / introduced
- Sep 15, 2025
- Public weights
- Public weights
- Licence / scope
- Not established
- Parameters
- Not disclosed
- Base model
- Qwen2.5-VL-3B
- Humanoid association
- None reported
- Inputs
- vision, language, proprioception
- Outputs
- robot action chunks
Release dates can mark an announcement or paper introduction. Public checkpoints, code licences and weight terms are distinct; follow the publisher’s links for the stated scope.
Disclosed training data
- AgiBot World →
- DROID →
- RH20T →
- BridgeData V2 →
- Self-collected robot trajectories, other open robot datasets and multimodal VQA
More robot AI models
Evidence
Sources
- [s1] Igniting VLMs toward the Embodied Space ↗
X Square Robot · paper · published Sep 15, 2025 · accessed Oct 13, 2026
- [s2] WALL-OSS flow model card and inference example ↗
X Square Robot · official · accessed Oct 13, 2026
- [s3] Wall-X official repository and model release branches ↗
X Square Robot · official · accessed Oct 13, 2026
- [s4] WALL-OSS paper: backbone and named training mixture ↗
X Square Robot · paper · published Sep 15, 2025 · accessed Oct 13, 2026