← Leaderboard

InternVLA-M1

VLA
Shanghai AI Lab · Oct 15, 2025

Spatially grounded dual-system VLA from InternRobotics: a VLM planner trained on spatial-grounding data guides a diffusion action expert. Weights appeared on Hugging Face in September 2025, a month before the paper.

Type
VLA
Parameters
4.1B
Backbone
Qwen2.5-VL-3B + DiT action expert
Weights
Open
License
CC BY-NC-SA 4.0
Commercial use
Not allowed
Released
Oct 15, 2025
Robots
Franka Research 3, WidowX (sim), Franka Panda (sim)
SO-101
–
LeRobot
–
Inputs
–
VRAM
–

Benchmark results

2 of 13 benchmarks
LIBEROSaturatedSimulation
#48 of 7095.9%, self-reported by the authors, not a leaderboard entry
SimplerEnv WidowXSaturatedSimulation
#11 of 3071.7%, self-reported by the authors, not a leaderboard entry
Not reported on 11 benchmarks
Wrong or missing number? Report an error