Interactive Demo

Every capability,
in your browser.

Pick a rollout from the sidebar and watch DreamZero-SO101 imagine the full trajectory from a single starting frame — across three camera views. Use the filters above to isolate one capability at a time.

Filter by capability
Request

Request a custom rollout

Send a scene + prompt. Live inference endpoint coming soon.

Open an issue

Suggest a scene via GitHub

Point us to an SO-101 episode on HF Hub and we will rollout it.

How to read the gallery

Eight capabilities on display

Use the filter chips above the gallery to isolate one capability at a time, or browse all 35 rollouts together.

01 — Simple pick

Single-object grasping

Pick a red, blue, yellow, or green cube/lego from a variety of scenes and camera rigs — across four community datasets.

02 — Pick and place

Container manipulation

Pick an object and place it in a bowl, bin, or transparent box. Tests trajectory planning beyond just grasping.

03 — Stacking

Two-object sequencing

Stack the red cube on the blue one — requiring approach, grasp, lift, translate, align, and release in imagination.

04 — Multi-object

Sequential manipulation

Pick all red cubes and place them in the bowl — the model must plan a multi-step sequence for several objects.

05 — Table cleanup

Four object categories

Pens, markers, tapes, and scissors — grab each and place into a pen holder. From so101-table-cleanup.

06 — Deformable

Cloth folding

Fold a cloth — the only deformable manipulation task in the training mix, from so101_cloth_folding1.

07 — Mid-episode

Resume from any state

Six rollouts that start deep inside a trajectory — descend, grasp, and lift phases of held-out episodes.

08 — Zero-shot

Out-of-distribution

Eight samples on so101_box_to_bowl_v2 — a dataset the model was never trained on.