OpenHelix team
VLA-Adapter
500M Commercial terms unverified 7/8 verified
1 of 8 required fields are unverified. They are shown as badges, not filled with defaults. Sources for verified fields are listed below.
Facts
- Released
- 2025-09 1 Source
- Parameters
- 0.5 B 1 Source
- Architecture
- VLM + action expert 0 Source
- License
- MIT (code repository); weights model card untagged 1 Source
- Commercial
- Commercial terms unverified 0 Source
- Weights
- huggingface.co ↗ 0 Source
- Code
- github.com ↗ 1 Source
- Paper
- arxiv.org ↗ 0 Source
- Input modalities
- rgb · language 0 Source
- Action space
- End-effector pose delta + gripper 0 Source
- Cross-embodiment
- LIBERO / CALVIN fine-tuning; ALOHA (Cobot Magic) real-robot deployment 1 Source
- Pretraining data (hours)
- Unverified 0 Source
- Target forms
- Single arm · Dual arm 0 Source
Architecture · Prismatic architecture + Qwen2.5-0.5B language backbone, lightweight policy head (Pro version policy approx. 207MB)
Sources · 1 sources
release_date,license,params_b,arch_notes,cross_embodiment,code_url https://github.com/OpenHelix-Team/VLA-Adapter Official page · Fetched 2026-09-04 · No snapshot stored yet README: paper 2025-09-11, code 2025-09-22; MIT license; 0.5B; Qwen2.5-0.5B; LIBERO / CALVIN; ALOHA / Cobot Magic
Activity signals
Public counters fetched daily from GitHub, Hugging Face and ModelScope. Numbers only; no judgement. · fetched 2026-10-31
GitHub stars
2,340
+21 / 30 d
Commits · 90 d
0
last push 227 d ago
HF downloads · 30 d
115
all-time 1,153 · 8 likes
Weights last modified
2025-09-21
406 d ago
ModelScopeVLA-Adapter/LIBERO-Spatial · 107 downloads · updated 2025-09-22 · same org as Hugging Face
Hugging FaceVLA-Adapter/LIBERO-Spatial
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 |
|---|---|---|---|
| CALVINVLA-Adapter | abc_d | 4.4 avg_len | paper · Table 6, row 'VLA-Adapter (Ours)', Avg. len |
| CALVINVLA-Adapter-Pro | abc_d | 4.5 avg_len | paper · Table 6, row 'VLA-Adapter-Pro (Ours)', Avg. len |
| LIBEROVLA-Adapter | avg | 97.3 | paper · Table 5, row 'VLA-Adapter (Ours)', column Avg. |
| LIBEROVLA-Adapter-Pro | avg | 98.5 | paper · Table 5, row 'VLA-Adapter-Pro (Ours)', column Avg. |
| LIBEROVLA-Adapter | goal | 97.2 | paper · Table 5, row 'VLA-Adapter (Ours)', column Goal |
| LIBEROVLA-Adapter-Pro | goal | 98.2 | paper · Table 5, row 'VLA-Adapter-Pro (Ours)', column Goal |
| LIBEROVLA-Adapter | long | 95 | paper · Table 5, row 'VLA-Adapter (Ours)', column Long |
| LIBEROVLA-Adapter-Pro | long | 96.4 | paper · Table 5, row 'VLA-Adapter-Pro (Ours)', column Long |
| LIBEROVLA-Adapter | object | 99.2 | paper · Table 5, row 'VLA-Adapter (Ours)', column Object |
| LIBEROVLA-Adapter-Pro | object | 99.6 | paper · Table 5, row 'VLA-Adapter-Pro (Ours)', column Object |
| LIBEROVLA-Adapter | spatial | 97.8 | paper · Table 5, row 'VLA-Adapter (Ours)', column Spatial |
| LIBEROVLA-Adapter-Pro | spatial | 99.6 | paper · Table 5, row 'VLA-Adapter-Pro (Ours)', column Spatial |
Deployment recipe
Not reproduced yet · Same as MiniVLA: depends on the prismatic runtime.
Record history
2026-09-04
Sources fetched
release_date,license,params_b,arch_notes,cross_embodiment,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 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| VLA-Adapter |
Legend Officially supportedCommunity reproducedFine-tunable in theoryUnsupportedUnknown Vendor-declared target form 3 / 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.