MolmoAct 2

MolmoAct 2 has 5.4B parameters. Estimated VRAM is 4.3 GB at Q4, 7.1 GB at Q8 and 13 GB at FP16. These figures are estimates, covering weights plus about 15% and 0.5 GB for a short context.

Yes. On a 24 GB or 16 GB GPU it fits at FP16 (estimated 13 GB), and on a 12 GB or 8 GB GPU it fits at Q8 (estimated 7.1 GB).

Yes, at Q4 the estimated VRAM is 4.3 GB. Assuming macOS lets the GPU use about 75% of unified memory, the smallest common Mac memory size that fits Q4 is 8 GB.

Yes, MolmoAct 2 has open weights, so you can self-host it for full control of your data.

Run MolmoAct 2 locally

It reads text and images. Audio, video and PDFs are not among its input modalities.

Closest alternatives, matched on model kind and release date with no comparable price, are Nex-N2-Pro by Nex AGI, MiMo-V2.5-Pro-UltraSpeed by Xiaomi and DiffusionGemma 26B A4B IT by Google DeepMind.

Nex-N2-ProMolmoAct 2 vs Nex-N2-ProMiMo-V2.5-Pro-UltraSpeedMolmoAct 2 vs MiMo-V2.5-Pro-UltraSpeedDiffusionGemma 26B A4B ITMolmoAct 2 vs DiffusionGemma 26B A4B IT

MolmoAct 2 is open weights.

The MolmoAct 2 weights are available in the Hugging Face repository allenai/MolmoAct2.

Run MolmoAct 2 locally

MolmoAct 2 was made by Ai2 and released on May 5, 2026.

More from Ai2

Yes, MolmoAct 2 has open weights.