Post-Cutoff
Search For AI agents

Ai2 (Allen Institute for AI)MolmoAct familyCurrentRobotics modelReleased 5 May 2026

MolmoAct 2 / MolmoAct 2-Think

Fully open (weights, data, code) VLA from Ai2, the main open alternative to π0.5/GR00T for tabletop manipulation. Start from a fine-tuned checkpoint (e.g. allenai/MolmoAct2-DROID) for ready-to-run inference; the base card has no inference code.

Model idallenai/MolmoAct2

PricePrice not published

Checked29 September 2026

Where it stands

11newer robotics models since its release

The 5 newest are listed. All robotics models

Weights
Weights are public. Licence: Apache-2.0; model weights on HF. See the licence for commercial terms.

How to call it

3 ways to use MolmoAct 2 / MolmoAct 2-Think, with the model id each provider expects.

Providers, model ids, endpoints and docs for MolmoAct 2 / MolmoAct 2-Think
ProviderModel idEndpointDocs
Hugging Face allenai/MolmoAct2 huggingface.co
GitHub github.com
Hugging Face LeRobot allenai/MolmoAct2-LIBERO-LeRobot huggingface.co

Pricing

Price not published.

No price source linked. Checked 29 September 2026.

Specs

Input
Text and images
Output
Robot actions
Context window
Not published
Max output
Not published
Open weights
Yes
Licence
Apache-2.0 (code); model weights on HF (license tag not stated on card)
Released
5 May 2026
Status
Current
Type
Robotics model

What stands out

What its maker marketed at launch, and what people found later, each with its source.

  • Open action reasoning model

    Molmo2-ER embodied-reasoning VLM connected to a flow-matching action expert via per-layer KV conditioning; the Think variant adds adaptive depth reasoning (interpretable depth map before acting).

    Source: allenai.org

  • Strong out-of-the-box real-world success

    87.1% average success over 15 real Franka tasks vs 45.2% for π0.5 and 48.4% for MolmoBot (Ai2’s evaluation); LIBERO 97.2% (98.1% Think).

    Source: allenai.org

  • Fast inference

    ~180 ms per action call (790 ms with adaptive depth reasoning) vs ~6,700 ms for the original MolmoAct (up to 37x faster).

    Source: allenai.org

  • Largest open bimanual dataset

    Released with MolmoAct2-BimanualYAM, 720+ hours of bimanual tabletop demonstrations, which Ai2 calls the largest open bimanual robotics dataset, plus an open FAST action tokenizer.

    Source: allenai.org

Notes

Checkpoints: MolmoAct2 (post-trained multi-embodiment foundation, ~5.4B params per HF safetensors), -Think, -Pretrain, fine-tuned -DROID, -BimanualYAM, -SO100_101, -LIBERO, -Think-LIBERO, FAST-Tokenizer. Main supported robots: SO-100/101, bimanual YAM, Franka (DROID); others need fine-tuning. Paper arXiv 2605.02881.

Sources: Ai2 blog, arXiv 2605.02881, HF model card, GitHub.