← Leaderboard
MolmoAct2
VLAAi2 · May 4, 2026
Ai2 action-reasoning model that pairs the Molmo2-ER vision-language model with a flow-matching action expert; it is loadable through the LeRobot bridge of the vla-eval harness.
- Type
- VLA
- Parameters
- 5.4B
- Backbone
- Molmo2-ER
- Weights
- Open
- License
- Apache-2.0
- Commercial use
- Unclear
- Released
- May 4, 2026
- Robots
- DROID Franka, Bimanual YAM, Franka Panda (sim)
- SO-101
- –
- LeRobot
- Supported
Benchmark results
3 of 13 benchmarksRoboDojo (sim)Simulation
- Official leaderboard ↗Sep 2, 2026
#31 of 405.0%
- Paper / model card ↗May 2026· n = 2000
#29 of 4697.2%, self-reported by the authors, not a leaderboard entry
SO-101 BenchLow-cost arm
- Paper / model card ↗Oct 27, 2026· n = 916
#9 of 1123.8%, self-reported by the authors, not a leaderboard entry
Not reported on 10 benchmarks
Adoption
updated Nov 17, 2026Running the model
| Hardware | Precision | Config | Latency / chunk | Steps | VRAM | Chunk | Control | Source |
|---|---|---|---|---|---|---|---|---|
| H100 | – | cache + CUDA Graphs | 179.2 ms | – | – | 10 | 10 Hz | May 4, 2026 ↗ |
| unspecified | fp32 | MolmoAct2-DROID checkpoint | – | – | 88 GB | – | – | Aug 23, 2026 ↗• |
Trained on
Wrong or missing number? Report an error