← Leaderboard

Pelican-VLA 0.5

VLA
X-Humanoid · Jul 7, 2026

X-Humanoid’s VLA built on its Pelican embodied VLM line. It self-reports 80% on a TienKung tabletop-cleanup task and RoboTwin 2.0 results with an adaptive execution strategy.

Type
VLA
Parameters
5B
Backbone
Qwen3-VL-4B + Cosmos tokenizer + flow-matching head
Weights
Open
License
Apache-2.0
Commercial use
Allowed
Released
Jul 7, 2026
Robots
TienKung, Aloha-AgileX (sim)
SO-101
–
LeRobot
Supported
Inputs
–
VRAM
–

Benchmark results

1 of 13 benchmarks
RoboTwin 2.0Near saturationSimulation
  • 91.2%, self-reported by the authors, not a leaderboard entryPaper / model card ↗Jul 2026· self-reported· non-standard
91.2%, self-reported by the authors, not a leaderboard entrynot ranked
Not reported on 12 benchmarks
Wrong or missing number? Report an error