Meta FAIR
V-JEPA 2-AC
1B 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
- 2025-06 1 Source
- Parameters
- 1 B 0 Source
- Architecture
- ViLLA 0 Source
- License
- Mostly MIT, some files Apache-2.0 (per README); HF weights require access request 1 Source
- Commercial
- Commercial terms unverified 0 Source
- Weights
- huggingface.co ↗ 1 Source
- Code
- github.com ↗ 1 Source
- Paper
- arxiv.org ↗ 0 Source
- Input modalities
- rgb · state 0 Source
- Action space
- End-effector pose (image-goal-based planning) 0 Source
- Cross-embodiment
- Reaching / grasping / pick-and-place planning on Franka 1 Source
- Pretraining data (hours)
- Unverified 0 Source
- Target forms
- Single arm 0 Source
Architecture · Action-conditioned predictor on top of the self-supervised video world model V-JEPA 2 (ViT-g 1B); zero-shot grasp/place via planning; weights are gated on HF
Sources · 2 sources
release_date,license,arch_notes,cross_embodiment,code_url https://github.com/facebookresearch/vjepa2 Official page · Fetched 2026-09-05 · No snapshot stored yet README：V-JEPA 2 2025-06-25；majority licensed under MIT, portions under separate terms；ViT-g/16 1B；Franka arm planning
weights_url https://huggingface.co/api/models/facebook/vjepa2-ac-vitg-384 Model hub page · Fetched 2026-09-05 · No snapshot stored yet Model card API returns 401 (gated access)
Activity signals
Public counters fetched daily from GitHub, Hugging Face and ModelScope. Numbers only; no judgement. · fetched 2026-10-31
GitHub stars
4,837
+129 / 30 d
Commits · 90 d
0
last push 3 d ago
HF downloads · 30 d
—
weights not on Hugging Face
Weights last modified
—
GitHubfacebookresearch/vjepa2 · 593 forks · 98 open issues · no GitHub release
ModelScopeno same-name mirror found (searched by repo name)
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) | grasp_box | 25 | paper · Table 2, row 'V-JEPA 2-AC (ours) / Avg', column 'grasp box' |
| Real-robot evaluation (self-reported) | grasp_cup | 65 | paper · Table 2, row 'V-JEPA 2-AC (ours) / Avg', column 'grasp cup' |
| Real-robot evaluation (self-reported) | pick_and_place_box | 65 | paper · Table 2, row 'V-JEPA 2-AC (ours) / Avg', column 'pick and place box' |
| Real-robot evaluation (self-reported) | pick_and_place_cup | 80 | paper · Table 2, row 'V-JEPA 2-AC (ours) / Avg', column 'pick and place cup' |
| Real-robot evaluation (self-reported) | reach | 100 | paper · Table 2, row 'V-JEPA 2-AC (ours) / Avg', column 'reach' |
| Real-robot evaluation (self-reported) | reach_with_object_box | 75 | paper · Table 2, row 'V-JEPA 2-AC (ours) / Avg', column 'reach with object box' |
| Real-robot evaluation (self-reported) | reach_with_object_cup | 75 | paper · Table 2, row 'V-JEPA 2-AC (ours) / Avg', column 'reach with object cup' |
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
release_date,license,arch_notes,cross_embodiment,code_url · weights_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 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| V-JEPA 2-AC |
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.