Skip to content
RobothecaBeta
Log in

InternVLA-M1

  • AI model
  • Humanoids and robot learning

A spatially guided vision-language-action model for generalist robot control, pairing a language head and an action head trained together.

  • Non-commercial weights

Made by Intern Robotics, Shanghai AI Laboratory.

Source on GitHub: github.com/InternRobotics/InternVLA-M1 (opens in a new tab). Facts read at commit 7f42cf2 (opens in a new tab).

Licenses

Code
MIT (opens in a new tab)

Read by Robotheca from LICENSE (opens in a new tab) (MIT) at commit 7f42cf2 (opens in a new tab), on 16 Oct 2026.

Commercial use
No

The weights license does not allow commercial use.

Labels

  • Non-commercial weightsThe weights license (CC-BY-NC-SA-4.0) does not allow commercial use.

What it does

  • Modular design where model architecture, training data, training strategies and evaluation are decoupled for independent development.
  • Dual-system design with a language head and an action head, trained with dual supervision under one framework.
  • Builds on open vision-language models such as Qwen2.5-VL and transfers spatial priors through spatial prompt fine-tuning.
  • Reports results on WindowX, Google Robot and LIBERO benchmarks, with higher scores than pi0 and GR00t in its comparison table.
  • Offers a chat demo for image question answering and spatial grounding, plus an action prediction demo that takes two camera views.
  • Provides examples for reproducing results in SimplerEnv and LIBERO, and for training and deployment on real robots, with several checkpoints on Hugging Face.

Written by Robotheca from the project's README (opens in a new tab) at commit 7f42cf2.

Getting started

Difficulty
Advanced
Hardware it needs, as the project states it
Runs on NVIDIA CUDA, Intel GPUs (XPU), or CPU; the scripts pick the available device in the order CUDA, XPU, then CPU. FlashAttention2 is installed for CUDA and skipped on XPU.

Maintenance

Status
Active: a commit in the last 90 days
Newest commit
9 Oct 2026 on InternVLA-M1
Latest release
None published on GitHub

Checked on GitHub on 16 Oct 2026.

Links

Robots on Robotheca that use it

No robot on Robotheca is linked to it yet. Links come from a robot's robot.yaml, from its dependency files read as text, or from Robotheca's editors.

Robotheca lists AI models as a reference: nothing is hosted or run here, and every fact links to where it was read.