Peking University EPIC Lab / Galbot
GraspVLA
Non-commercial 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
- 2025-07 1 Source
- Parameters
- Unverified 0 Source
- Architecture
- VLM + action expert 0 Source
- License
- CC BY-NC 4.0 (code badge and model card agree) 2 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 0 Source
- Action space
- Grasp pose / end-effector trajectory 0 Source
- Cross-embodiment
- Synthetic data SynGrasp-1B (240 categories, 10k+ objects) 1 Source
- Pretraining data (hours)
- Unverified 0 Source
- Target forms
- Single arm 0 Source
Architecture · InternLM2-1.8B language + DINOv2/SigLIP vision; pretrained on one billion frames of synthetic grasp data SynGrasp-1B
Sources · 2 sources
release_date,license,arch_notes,cross_embodiment,code_url https://github.com/PKU-EPIC/GraspVLA Official page · Fetched 2026-09-04 · No snapshot stored yet README: model released 2025-07-25; CC BY-NC 4.0; internlm2-1_8b + DINOv2/SigLIP; SynGrasp-1B
weights_url,license https://huggingface.co/api/models/vegebirrd/GraspVLA Model hub page · Fetched 2026-09-04 · No snapshot stored yet Model card license: cc-by-nc-4.0
Activity signals
Public counters fetched daily from GitHub, Hugging Face and ModelScope. Numbers only; no judgement. · fetched 2026-10-31
GitHub stars
426
+7 / 30 d
Commits · 90 d
2
last push 74 d ago
HF downloads · 30 d
109
all-time 231 · 1 likes
Weights last modified
2026-08-18
75 d ago
GitHubPKU-EPIC/GraspVLA · 19 forks · 1 open issues · no GitHub release
ModelScopeno same-name mirror found (searched by repo name)
Hugging Facevegebirrd/GraspVLA
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 |
|---|---|---|---|
| LIBEROGraspVLA (zero-shot, grasp-only LIBERO) | goal | 91.2 | paper · Table 2, row 'Ours (zero-shot)', column Goal |
| LIBEROGraspVLA (zero-shot, grasp-only LIBERO) | long | 82 | paper · Table 2, row 'Ours (zero-shot)', column Long |
| LIBEROGraspVLA (zero-shot, grasp-only LIBERO) | object | 94.1 | paper · Table 2, row 'Ours (zero-shot)', column Object |
| Other benchmarks | Language-Conditioned overall | 93.3 | paper · Table 3, row 'Ours', Language-Conditioned overall |
| Real-robot evaluation (self-reported) | Synthetic Categories / b.g. | 93.3 | paper · Table 1, row 'Ours', Synthetic Categories, column 'b.g.' |
| Real-robot evaluation (self-reported) | Synthetic Categories / basic | 93.3 | paper · Table 1, row 'Ours', Synthetic Categories, column 'basic' |
| Real-robot evaluation (self-reported) | Synthetic Categories / dis. | 93.3 | paper · Table 1, row 'Ours', Synthetic Categories, column 'dis.' |
| Real-robot evaluation (self-reported) | Synthetic Categories / height | 90 | paper · Table 1, row 'Ours', Synthetic Categories, column 'height' |
| Real-robot evaluation (self-reported) | Synthetic Categories / light | 96.6 | paper · Table 1, row 'Ours', Synthetic Categories, column 'light' |
| Real-robot evaluation (self-reported) | Web Categories / b.g. | 93.3 | paper · Table 1, row 'Ours', Web Categories, column 'b.g.' |
| Real-robot evaluation (self-reported) | Web Categories / basic | 93.3 | paper · Table 1, row 'Ours', Web Categories, column 'basic' |
| Real-robot evaluation (self-reported) | Web Categories / dis. | 86.6 | paper · Table 1, row 'Ours', Web Categories, column 'dis.' |
| Real-robot evaluation (self-reported) | Web Categories / height | 86.6 | paper · Table 1, row 'Ours', Web Categories, column 'height' |
| Real-robot evaluation (self-reported) | Web Categories / light | 90 | paper · Table 1, row 'Ours', Web Categories, column 'light' |
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
release_date,license,arch_notes,cross_embodiment,code_url · weights_url,license
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 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| GraspVLA |
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.