MolmoBot
MolmoBot is a large-scale, fully-simulated robot manipulation dataset from Ai2 (Allen Institute for AI), built to enable zero-shot sim-to-real transfer.
Source notes
MolmoBot is a large-scale, fully-simulated robot manipulation dataset from Ai2 (Allen Institute for AI), built to enable zero-shot sim-to-real transfer.
- ~1.7M expert trajectories (5,704 hours) across 8 task types — articulated-object manipulation (opening doors, drawers, cabinets) and pick-and-place
- Generated entirely in simulation (MuJoCo via Ai2 MolmoSpaces) with aggressive domain randomization over objects, placements, viewpoints, lighting, textures, and dynamics
- 94K+ procedurally-generated environments, 11K+ unique objects, 9K+ receptacles
- Two robot platforms: Franka FR3 (DROID tabletop) and Rainbow Robotics RB-Y1 (mobile bimanual)
- Trained purely on this simulated data, MolmoBot reaches 79.2% zero-shot real-world tabletop pick-and-place success, vs 39.2% for pi-0.5
- Distributed as Parquet (auto-converted from tar.zst); ~10.3 TB. License: ODC-BY 1.0 (Ai2 Responsible Use Guidelines).
- Scale
- 5,704 hours
- Formats
- custom
- License
- custom
- Published
- 2026-03-17
Decision summary
scripted
Metadata requires review against the official source before publication.
Read the dataset manifest and feature schema.
Release history
Source-backed release timing for this canonical dataset record.
- Release evidence
Dataset series published
Release timing is recorded from official dataset metadata.
Selection readiness
6 evidence dimensions for deciding whether this dataset is ready to inspect, compare, or adopt. This is not a model benchmark.
Catalog evidence · not task performance
How scores work70/100
Provisional
3/6 dimensions scored · 42% confidence
Read the evidence behind all 6 dimensions
Access and governance
Access and license are declared by the source.
Schema and signal coverage
No machine-readable schema has been verified yet.
Policy training readiness
Observation and action/state signals are declared; alignment quality still depends on sample verification.
World-model readiness
World-model observation, geometry, or temporal semantics are not verified.
Failure and recovery readiness
No verified failure/recovery annotation evidence is available yet.
Download and processing readiness
Scale is declared; transfer and processing estimates are not measured.
Review unresolved evidence and next checks
Signal gaps
- Gaze / attention. Not enough evidence is available to classify this signal.
- Feedback / correction / failure. Not enough evidence is available to classify this signal.
Next checks
- Read the dataset manifest and feature schema.
- Run a bounded sample audit before assigning Strong readiness.
Engineering checks (OBRS) behind this record
Needs Audit
OBRS metadata and reported test evidence. A breakdown behind Selection readiness, not a separate score or an independent OpenBot certification.
- A passing real-hardware test report is required.
- A passing ingestion/pipeline test report is required.
- A passing privacy and provenance review is required.
Dataset facts
- Source
- Allen Institute for AI (Ai2)
- Evidence
- secondary claim
- Formats
- custom
- episodes
- 1.7M
- hours
- 5,704
- tasks
- 8
Loop signals
5/7 present or partial
No decision-grade evidence captured yet.
unknownNo decision-grade evidence captured yet.
unknownView 5 more signal categories
video
presentee_pose · **MolmoBot** is a large-scale, **fully-simulated** robot manipulation dataset from Ai2 (Allen Institute for AI), built to enable **zero-shot sim-to-real** transfer. - **~1.7M expert trajectories** (**5,704 hours**) across **8 task types** — articulated-object manipulation (opening doors, drawers, cabinets) and pick-and-place - Generated entirely in simulation (**MuJoCo** via Ai2 **MolmoSpaces**) with aggressive domain randomization over objects, placements, viewpoints, lighting, textures, and dynamics - **94K+ procedurally-generated environments**, **11K+ unique objects**, **9K+ receptacles** - Two robot platforms: **Franka FR3** (DROID tabletop) and **Rainbow Robotics RB-Y1** (mobile bimanual) - Trained purely on this simulated data, MolmoBot reaches **79.2%** zero-shot real-world tabletop pick-and-place success, vs 39.2% for pi-0.5 - Distributed as Parquet (auto-converted from tar.zst); ~10.3 TB. License: **ODC-BY 1.0** (Ai2 Responsible Use Guidelines).
partial**MolmoBot** is a large-scale, **fully-simulated** robot manipulation dataset from Ai2 (Allen Institute for AI), built to enable **zero-shot sim-to-real** transfer. - **~1.7M expert trajectories** (**5,704 hours**) across **8 task types** — articulated-object manipulation (opening doors, drawers, cabinets) and pick-and-place - Generated entirely in simulation (**MuJoCo** via Ai2 **MolmoSpaces**) with aggressive domain randomization over objects, placements, viewpoints, lighting, textures, and dynamics - **94K+ procedurally-generated environments**, **11K+ unique objects**, **9K+ receptacles** - Two robot platforms: **Franka FR3** (DROID tabletop) and **Rainbow Robotics RB-Y1** (mobile bimanual) - Trained purely on this simulated data, MolmoBot reaches **79.2%** zero-shot real-world tabletop pick-and-place success, vs 39.2% for pi-0.5 - Distributed as Parquet (auto-converted from tar.zst); ~10.3 TB. License: **ODC-BY 1.0** (Ai2 Responsible Use Guidelines).
partialsimulation · **MolmoBot** is a large-scale, **fully-simulated** robot manipulation dataset from Ai2 (Allen Institute for AI), built to enable **zero-shot sim-to-real** transfer. - **~1.7M expert trajectories** (**5,704 hours**) across **8 task types** — articulated-object manipulation (opening doors, drawers, cabinets) and pick-and-place - Generated entirely in simulation (**MuJoCo** via Ai2 **MolmoSpaces**) with aggressive domain randomization over objects, placements, viewpoints, lighting, textures, and dynamics - **94K+ procedurally-generated environments**, **11K+ unique objects**, **9K+ receptacles** - Two robot platforms: **Franka FR3** (DROID tabletop) and **Rainbow Robotics RB-Y1** (mobile bimanual) - Trained purely on this simulated data, MolmoBot reaches **79.2%** zero-shot real-world tabletop pick-and-place success, vs 39.2% for pi-0.5 - Distributed as Parquet (auto-converted from tar.zst); ~10.3 TB. License: **ODC-BY 1.0** (Ai2 Responsible Use Guidelines).
presentOpen · custom · custom
presentEvidence details and provenance
Official signal claims
Schema and annotations
No machine-readable schema facts are captured.
Sample verification
Pending. Metadata does not prove sample coverage, alignment, or file integrity.
curated official source
Integration notes
- Metadata requires review against the official source before publication.