Topics / #vla

Vla projects on GitHub

Vision-Language-Action models that map visual and text input to robot actions

As of , the most-starred Vla projects on GitHub are Zetta-Embodiment (1.3K stars), OpenWAM (1K stars) and LoongForge (687 stars). GitHub.Locker tracks 9 Vla repositories with 4.6K stars combined. The fastest-growing on October 16 was OpenWAM, with +2 stars in a day.

9Projects
4.6KStars in total
+6Stars in a day
Browse in the catalog →github.com/topics/vla ↗

Hot right now

Most stars gained in a day · Oct 16

+2

OpenWAM

OpenWAM-Official/OpenWAM · ★ 1K

OpenWAM is an open research stack for systematically building World-Action Models (WAMs) for robots, breaking the usual tightly-coupled design choices into modular, swappable components you can compare in controlled experiments.

+2

Zero-WAM

robbyant-research/Zero-WAM · ★ 434

Zero-WAM is a research method for zero-shot cross-task robotic manipulation: a causal video-action policy that learns in-context world-action modeling, predicting future robot observations and executable actions from either language instructions or human demonstration videos.

+1

RoboICL

Mosi-AI/RoboICL · ★ 268

RoboICL is a research framework that lets a frozen, general-purpose vision-language model (GPT-6 Astra) control a robot in simulation purely through in-context learning, without training a dedicated robot policy.

+1

flex-pi

geyan21/flex-pi · ★ 225

Flex-π is a 6B-parameter world-action model for robot manipulation that jointly predicts future RGB, 3D pointmaps, DINOv3 semantics, and actions from one checkpoint, so it can deploy as a VLA, a full world model, or anything in between.

The all-time top

9 projects · by stars

  1. #1 · air-embodied-brain/Zetta-Embodiment

    Zetta-Embodiment

    Zetta is a closed-loop embodied harness that evolves code-based runtime critics and recovery skills online while keeping the base robot policy frozen, reaching 90.8% success on LIBERO-Pro and 93.6% on RoboCasa with an 11.1x inference speedup.

    1.3K
  2. #2 · OpenWAM-Official/OpenWAM

    OpenWAM

    OpenWAM is an open research stack for systematically building World-Action Models (WAMs) for robots, breaking the usual tightly-coupled design choices into modular, swappable components you can compare in controlled experiments.

    1K
  3. #3 · baidu-baige/LoongForge

    LoongForge

    An open-source training framework from Baidu's Baige team for LLMs, VLMs, diffusion and embodied models, built for faster training and lower cost on NVIDIA GPUs and Kunlun XPUs.

    687
  1. 4

    Zero-WAM

    434robbyant-research/Zero-WAM · Python

    Zero-WAM is a research method for zero-shot cross-task robotic manipulation: a causal video-action policy that learns in-context world-action modeling, predicting future robot observations and executable actions from either language instructions or human demonstration videos.

  2. 5

    RoboICL

    268Mosi-AI/RoboICL · Python

    RoboICL is a research framework that lets a frozen, general-purpose vision-language model (GPT-6 Astra) control a robot in simulation purely through in-context learning, without training a dedicated robot policy.

  3. 6

    quackd

    263rokbenko/quackd · Python

    quackd is an open-source daemon that gives small robots a brain: it takes a goal in plain language and plans the right sequence of the robot's own existing skills to carry it out, using an LLM like Claude, OpenAI, Gemini, Grok or a local model.

  4. 7

    flex-pi

    225geyan21/flex-pi · Python

    Flex-π is a 6B-parameter world-action model for robot manipulation that jointly predicts future RGB, 3D pointmaps, DINOv3 semantics, and actions from one checkpoint, so it can deploy as a VLA, a full world model, or anything in between.

  5. 8

    StreamPI

    211hku-sail/StreamPI · Python

    Official code of StreamPI (NeurIPS 2026, from HKU and ACE Robotics): a streaming temporal modeling framework that gives VLA robot policies like pi0.5 memory of past frames and multi-frame perception without adding any new parameters.

  6. 9

    dreamforge

    130rzc156/dreamforge · Python

    Dreamforge is a research pipeline that mines a robot policy's own failure states, uses a latent world model (CEM) to imagine corrective actions, and verifies each correction by replaying it in the real simulator before claiming it helps.

Often tagged together