robot policy
OpenVLA-OFT
Fine-tuning recipe for OpenVLA reaching 97.1% average on LIBERO.
Levelmodel
Classrobot policy
Parameters7B
Contextnot in the notes
Released27 Feb 2025
Opennessopen source
Runs onnot in the notes
Statusactive
Benchmark Results
vendor-reported
LIBEROworld models and robotics
Averagevendor-reported
97.1
050100
Score in %. Higher is better.
OpenVLA-OFT paper
Where this comes from
Every score names who ran it: vendor-reported, maintainer-run or independent. A rank is shown with the population it was ranked in (gap rule 3).
Lineage
finetune
OpenVLA is the parent (finetune)high confidencepress claimWhere this comes from
Parent and child relations from model cards, Hugging Face base_model fields and the genealogy tables of the research notes. The tree shows two generations each way, 3 models. Confidence and evidence are the notes' own.
Identity
- Developer
- Stanford University
- Family
- OpenVLA
- Released
- 27 Feb 2025 sourcesregistry date
- Weights published
- –
- Record checked
- 10 Oct 2026
Where this comes from
Developer and dates from the research notes. Release date is the vendor's public announcement; weights date is when open weights were published.
Architecture and Size
- Architecture
- not in the notes Optimised fine-tuning recipe: parallel decoding, action chunking, continuous actions
- Total parameters
- 7B sources
- Active parameters
- –
- Checkpoint parameters
- –
- Context length
- not in the notes
- Input
- image, text
- Output
- action
Where this comes from
Parameter counts as the developer states them. Where the checkpoint holds a different total, both are shown (register A16).
Openness and Access
- Openness
- open source
- Licence
- not in the notes sources
- Access routes
- open weights download
Where this comes from
The licence file governs over registry tags and aggregator labels.
Deployment
- Runs on
- not in the notes
- Footprint
- not in the notes
- Runtimes
- not in the notes
Where this comes from
Deployment tiers and runtimes as named in the notes. Footprint is the notes' figure, not a measurement by AURION.
Indonesia
- Indonesian support
- not stated
- Regional languages
- not in the notes
- Jakarta hosting
- unknown
- Languages
- not in the notes
Where this comes from
Indonesia fields are carried on every model; unknown is an allowed value.
Industry Use Cases
No industry use case links this model.
Links
DeveloperStanford Universityuniversity, USA
Stanford Universityuniversity, USAModelOpenVLA-OFTStanford University, robot policy
OpenVLA-OFTStanford University, robot policyR&D TopicsVision-Language-Action Robot Policies
Use CasesHumanoid Parts Handling in Automotive ManufacturingIndustrialRetail Grocery Manipulation RobotsRetail and LogisticsHumanoid Robots in LogisticsRetail and LogisticsUse cases of models in the same R&D topic; none is recorded for this model itself.
Who Runs It
BMW (Spartanburg plant)In production, 30 Jun 2026Catalyst BrandsContract signed, 26 May 2026Deployments of these use cases; none names this model.
BMW (Spartanburg plant)In production, 30 Jun 2026Catalyst BrandsContract signed, 26 May 2026Deployments of these use cases; none names this model.Models 2
Organisations 1
Sources
- Fine-Tuning Vision-Language-Action Models: Optimizing Speed and Success arxiv.org https://arxiv.org/abs/2502.19645
To cite this page: Project AURION, AI Use-Cases Research & Innovation, "OpenVLA-OFT", https://aurion.semai.id/models/stanford-university-openvla-oft/, records of October 2026.