Tsinghua University TSAIL (thu-ml)
RDT-1B
1B Commercial use OK 8/8 verified
Facts
- Released
- 2024-10 1 Source
- Parameters
- 1 B 1 Source
- Architecture
- Transformer policy 0 Source
- License
- MIT (code, weights and data all MIT, per README) 2 Source
- Commercial
- Commercial use OK 0 Source
- Weights
- huggingface.co ↗ 1 Source
- Code
- github.com ↗ 1 Source
- Paper
- arxiv.org ↗ 1 Source
- Input modalities
- rgb · language · state 0 Source
- Action space
- Unified action space covering single-arm/dual-arm, joints/end-effector, position/velocity 0 Source
- Cross-embodiment
- Pretrained on 1000k+ multi-robot episodes, fine-tuned on 6K+ ALOHA dual-arm episodes 1 Source
- Pretraining data (hours)
- Unverified 0 Source
- Target forms
- Single arm · Dual arm · Mobile manipulator 0 Source
Architecture · Diffusion Transformer; inputs language + up to three RGB streams, predicts 64 action steps at once; smaller RDT-170M version also available
Sources · 3 sources
license,params_b,cross_embodiment,code_url https://github.com/thu-ml/RoboticsDiffusionTransformer Official page · Fetched 2026-09-04 · No snapshot stored yet README：All the code, model weights, and data are licensed under MIT license；1B-parameter；1M+ multi-robot episodes
weights_url,license https://huggingface.co/api/models/robotics-diffusion-transformer/rdt-1b Model hub page · Fetched 2026-09-04 · No snapshot stored yet Model card license: mit; createdAt 2024-08-27
paper_url,release_date https://arxiv.org/abs/2410.07864 Paper / official docs · Fetched 2026-09-04 · No snapshot stored yet RDT-1B: a Diffusion Foundation Model for Bimanual Manipulation，2024-10
Activity signals
Public counters fetched daily from GitHub, Hugging Face and ModelScope. Numbers only; no judgement. · fetched 2026-10-31
GitHub stars
1,814
+8 / 30 d
Commits · 90 d
0
last push 284 d ago
HF downloads · 30 d
277
all-time 35,869 · 106 likes
Weights last modified
2024-10-17
745 d ago
GitHubthu-ml/RoboticsDiffusionTransformer · 170 forks · 42 open issues · no GitHub releaseno push for 283 d
ModelScopewardvecker/rdt-1b · 1 downloads · updated 2026-08-14 · same-name mirror, not the vendor org
Hugging Facerobotics-diffusion-transformer/rdt-1b
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 |
|---|---|---|---|
| Real-robot evaluation (self-reported) | agibot_g1_5_tasks_avg_completion_score | 0.5 score 0-1 | third-party · AgiBot World Colosseo / GO-1 paper, Fig. 5(b) 'Average' bar label for RDT-1B |
| Real-robot evaluation (self-reported) | aloha_fold_shorts_1shot_total | 68 | paper · Table 3, row 'RDT (ours)', Fold Shorts (1-Shot) Total |
| Real-robot evaluation (self-reported) | aloha_handover_5shot_total | 40 | paper · Table 3, row 'RDT (ours)', Handover (5-Shot) Total (sub-steps Pick Up Pen 100, Switch Hand 56, Drop Pen 56, Fall into Box 40) |
| Real-robot evaluation (self-reported) | aloha_pour_water_L_1/3_total | 100 | paper · Table 3, row 'RDT (ours)', Pour Water-L-1/3 'Total' (instruction following); Correct Hand 100, Correct Amount 100 |
| Real-robot evaluation (self-reported) | aloha_pour_water_unseen_room3_total | 62.5 | paper · Table 3, row 'RDT (ours)', Pour Water 'Total' sub-column for unseen room 3 (used as 'Unseen Scene' in Table 2) |
| Real-robot evaluation (self-reported) | aloha_robot_dog_total | 76 | paper · Table 3, row 'RDT (ours)', Robot Dog (Dexterity) 'Total' column (Grab Remote 100, Push Joystick 76, Total 76, Walk Straight 48) |
| Real-robot evaluation (self-reported) | aloha_wash_cup_unseen_cup2_total | 50 | paper · Table 3 'Quantitative results', row 'RDT (ours)', Wash Cup 'Total' sub-column for unseen cup 2 (also used as 'Unseen Object' in Table 2) |
| RoboTwin 2.0 | easy | 34.5 | third-party · X-VLA paper, Table 2, row 'RDT (Liu et al., 2025b)', RoboTwin-2.0 Easy |
| RoboTwin 2.0 | hard | 13.7 | third-party · X-VLA paper, Table 2, row 'RDT (Liu et al., 2025b)', RoboTwin-2.0 Hard |
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
2026-09-04
Sources fetched
license,params_b,cross_embodiment,code_url · weights_url,license · paper_url,release_date
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 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| RDT-1B |
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.