NVIDIA
GR00T N1.5 3B
2.72B Non-commercial 8/8 verified
Facts
- Released
- 2025-05 1 Source
- Parameters
- 2.72 B 1 Source
- Architecture
- Dual system 0 Source
- License
- Model card tagged Nvidia License (page states non-commercial use only) 1 Source
- Commercial
- Non-commercial 0 Source
- Weights
- huggingface.co ↗ 1 Source
- Code
- github.com ↗ 1 Source
- Paper
- arxiv.org ↗ 0 Source
- Input modalities
- rgb · language · state 0 Source
- Action space
- Joints (flow matching action chunks) 0 Source
- Cross-embodiment
- Cross-embodiment model, humanoid / dual-arm / single-arm (model card lists no specific robots) 0 Source
- Pretraining data (hours)
- Unverified 0 Source
- Target forms
- Humanoid · Dual arm · Single arm 0 Source
Architecture · SigLIP2 vision + T5 language encoder + flow matching action Transformer (adaptive layernorm); variable-dimension proprioception indexed by embodiment ID; superseded by N1.6 / N1.7
Sources · 3 sources
params_b,release_date,weights_url https://huggingface.co/api/models/nvidia/GR00T-N1.5-3B Model hub page · Fetched 2026-09-05 · No snapshot stored yet safetensors 2,724,163,520；createdAt 2025-05-28
license,arch_notes https://huggingface.co/nvidia/GR00T-N1.5-3B Model hub page · Fetched 2026-09-05 · No snapshot stored yet Model card: Nvidia License (Non-commercial use only); SigLip2 + T5 + flow matching transformer; cross-embodiment
code_url https://github.com/NVIDIA/Isaac-GR00T Official page · Fetched 2026-09-05 · No snapshot stored yet Isaac-GR00T code repository (Apache-2.0)
Activity signals
Public counters fetched daily from GitHub, Hugging Face and ModelScope. Numbers only; no judgement. · fetched 2026-10-31
GitHub stars
8,296
+146 / 30 d
Commits · 90 d
11
last push 3 d ago
HF downloads · 30 d
1,590
all-time 79,508 · 199 likes
Weights last modified
2025-09-17
410 d ago
GitHubNVIDIA/Isaac-GR00T · 1,515 forks · 345 open issues · latest release n1.7-release (2026-04-18)
ModelScopenv-community/GR00T-N1.5-3B · 748 downloads · updated 2025-09-18 · same-name mirror, not the vendor org
Hugging Facenvidia/GR00T-N1.5-3B
stars · last 30 d
Public benchmark scores
As printed in the source; protocols differ between papers. Full table with notes on the benchmarks page.
| Benchmark | Subset | Score | Source |
|---|---|---|---|
| Other benchmarks | 0_shot | 43.9 | official · 'Data-Limited Post-Training' table, row 'Sim GR-1, 0-shot', GR00T N1.5 (N1 = 39.6); row 'Sim GR-1, 30 Demos per Task' = 47.4 (N1 = 43.2) |
| Other benchmarks | success_rate | 38.3 | official · 'DreamGen Tasks (12 new verbs)' section: GR00T N1.5 38.3% (N1 13.1%) |
| Real-robot evaluation (self-reported) | real_gr1_language_following_task_overall_success | 83 | official · 'Real GR-1 Language Following Task' table, 'Overall success rate', GR00T N1.5 (N1 = 43.3%); 'Language following rate' 93.3% (N1 = 46.6%) |
| Real-robot evaluation (self-reported)GR00T N1.5 post-trained (1K demos) | unitree_g1_place_1_of_2_fruits | 98.8 | official · 'Unitree G1 Post-training (1K demos)' table, row 'GR00T N1.5, 1K Demos', 'Place 1 of 2 fruits (4 total)' (N1 = 44.0%); 'Place 1 of 2 objects (5 novel)' = 84.2% |
| RoboCasaGR00T N1.5 post-trained (30 demos/task) | 30_demos_per_task | 47.5 | official · GR00T N1.5 research page, 'Data-Limited Post-Training' table, row 'RoboCasa, 30 Demos per Task', column GR00T N1.5 (GR00T N1 = 17.4) |
Deployment recipe
No recipe yet. We publish only steps that actually ran on a rented GPU with a measured latency; this model has not been through that yet.
Record history
2026-09-05
Sources fetched
params_b,release_date,weights_url · license,arch_notes · code_url
Latency & cost
cost = p50 ms ÷ 1000 × rental ÷ 3600 × 1000
| Hardware | p50 ms | p95 ms | Peak VRAM | Cost per 1k inferences ¥ | samples |
|---|---|---|---|---|---|
NVIDIA RTX 4090 ¥1.68/h · GongJi Compute (2026) | — | — | — | — no latency data | No data |
NVIDIA RTX 5090 ¥2.4/h · Runyun 2.4 / GongJi 2.5 / Paratera Cloud 2.98 (2026) | — | — | — | — no latency data | No data |
NVIDIA A100 80GB no rental price | — | — | — | — no latency data | No data |
NVIDIA Jetson AGX Thor no rental price | — | — | — | — no latency data | No data |
NVIDIA H100 80GB no rental price | — | — | — | — no latency data | No data |
NVIDIA A800 80GB no rental price | — | — | — | — no latency data | No data |
No latency measurements for this model yet · Data comes from literature, maintainer runs, or crowd CLI uploads. None exist so far.
Compatibility
Colored only where we hold evidence. An amber dot means the vendor declares this form as a target; that is a claim, not a verified compatibility.
| Unitree G1 | Unitree H1 | AgiBot Genie G1 | AgileX Piper arm | SO-101 | Franka FR3 / Panda | Fourier GR-2 | WidowX 250 S | Google Robot（Everyday Robots） | ALOHA 2 / Cobot Magic (dual-arm) | Galaxea R1 Pro | Franka Emika Panda | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| GR00T N1.5 3B |
Legend Officially supportedCommunity reproducedFine-tunable in theoryUnsupportedUnknown Vendor-declared target form 1 / 12 cells with a definite status
Reproduction records
Did this model run on your robot? Report it with a link (log, video or repo). Records are reviewed by the maintainers; a verified “works” or “works after fine-tuning” upgrades the matrix cell to “community verified” with your link as evidence. Anonymous is fine; nothing but the fields below is stored.
No verified record yet for this model.
Clear direction, grounded judgement.