# Meta released V-JEPA 2 and physical-reasoning benchmarks for video world-model research

Meta released V-JEPA 2 on June 11, 2025, pairing a video-trained world model with physical-reasoning benchmarks and reported zero-shot robot-planning experiments.

Canonical: https://rangoon.ai/insights/archive-meta-v-jepa-2-june-2025/
Author: Rangoon Editorial (https://rangoon.ai/insights/editorial/)
Published: 2026-10-24
Updated: 2026-10-24
Event date: 2025-06-11
Tags: [Meta AI](https://rangoon.ai/insights/tags/meta-ai/), [V-JEPA 2](https://rangoon.ai/insights/tags/v-jepa-2/), [world models](https://rangoon.ai/insights/tags/world-models/), [robotics research](https://rangoon.ai/insights/tags/robotics-research/)
Source note: Historical reporting: Meta introduced V-JEPA 2 on June 11, 2025 and released code, checkpoints and three benchmarks. The blog and same-day research publication describe video pre-training, action-conditioned post-training and laboratory robot experiments. Meta’s benchmark and robot results are research results under stated conditions, not evidence of general robot reliability, safety certification or a production deployment.

Image: Video-to-action research stages. Original Rangoon explanatory diagram: Source boundary: dated primary release reporting; original editorial analysis. No vendor interface, photograph, measurement result, or endorsement implied.

## Key takeaways
- V-JEPA 2 separated broad video learning from a later action-conditioned stage intended to support planning experiments.
- Meta released benchmarks alongside the model, making reported physical-reasoning results conditional on defined tasks and metrics.
- Laboratory zero-shot robot demonstrations did not establish certification, broad operational reliability or safe deployment in uncontrolled settings.

## The June event joined a video world model to three physical-reasoning benchmarks [source 1](https://ai.meta.com/blog/v-jepa-2-world-model-benchmarks/) [source 2](https://ai.meta.com/research/publications/v-jepa-2-self-supervised-video-models-enable-understanding-prediction-and-planning/)

On June 11, 2025, Meta introduced V-JEPA 2 as a video-trained world model and said it was releasing three benchmarks for physical reasoning. The launch describes the research goal as learning from observations, predicting how a scene may evolve and using those predictions to plan actions. It also says code and model checkpoints were available for commercial and research applications. That is a public research-artifact event. It does not show that a model has a complete understanding of a physical environment or that a planned action will be safe in a new setting.
Meta describes the system as an encoder that maps raw video to semantic embeddings and a predictor that produces expected embeddings from context. The distinction matters because the release is built around learned representations and predicted consequences rather than a direct image-to-command guarantee. A benchmark result may show progress on a defined understanding, prediction or planning task, while leaving sensors, object properties, control hardware, latency and human oversight outside the result. The June record supports a technical direction, not an operational promise.


## The research record separated action-free learning from later action-conditioned work [source 1](https://ai.meta.com/blog/v-jepa-2-world-model-benchmarks/) [source 2](https://ai.meta.com/research/publications/v-jepa-2-self-supervised-video-models-enable-understanding-prediction-and-planning/)

Meta said V-JEPA 2 first used self-supervised, action-free pre-training on more than one million hours of video and one million images. It then added action-conditioned training with robot data so the predictor could account for candidate actions. This two-stage structure is important for interpreting the claim. Broad video learning can support motion and scene representations, while a later control-oriented stage gives the system information about actions. Neither stage alone turns a video model into a generally capable robot operator.
The publication says the action-conditioned version, V-JEPA 2-AC, used less than 62 hours of unlabeled robot video from the DROID dataset. It reports that the model was deployed zero-shot on Franka arms in two labs for image-goal picking and placing without collecting data in those environments. Those are bounded laboratory experiments with named hardware, tasks and goal formats. They should not be widened into a finding about homes, factories, public spaces, people nearby or systems that must recover safely from unseen failures.


## Reported scores belonged to their datasets, readouts and test conditions [source 1](https://ai.meta.com/blog/v-jepa-2-world-model-benchmarks/) [source 2](https://ai.meta.com/research/publications/v-jepa-2-self-supervised-video-models-enable-understanding-prediction-and-planning/)

Meta reported motion-understanding, action-anticipation and video-question-answering results, including measurements on Something-Something v2, Epic-Kitchens-100, PerceptionTest and TempCompass. The research page provides the numerical scores and says some video-question-answering results followed alignment with a language model. Those facts are useful for locating the work in a published evaluation record. They are not portable rankings for every video task, because data distribution, readout method, prompt design and comparison system all affect the measured result.


## World-model research did not amount to a certified physical agent [source 1](https://ai.meta.com/blog/v-jepa-2-world-model-benchmarks/) [source 2](https://ai.meta.com/research/publications/v-jepa-2-self-supervised-video-models-enable-understanding-prediction-and-planning/)

This article does not project later robot trials, model updates or field deployments back into June 2025. It treats Meta’s results as vendor-published research under the methods described in the source records. The historical conclusion is narrower: Meta released V-JEPA 2 with a staged video-and-action learning design, reported controlled robot-planning experiments and supplied benchmarks for further physical-reasoning evaluation. Those facts support scrutiny and replication, not a claim that physical agents were ready for unrestricted operation.


## Sources
1. [Meta AI: Introducing the V-JEPA 2 world model and new benchmarks for physical reasoning](https://ai.meta.com/blog/v-jepa-2-world-model-benchmarks/)
2. [Meta AI Research: V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and Planning](https://ai.meta.com/research/publications/v-jepa-2-self-supervised-video-models-enable-understanding-prediction-and-planning/)

## Related reading
- [Wayve launched GAIA-2 for controllable multi-camera driving scenarios, positioning it as an offboard training and validation tool](https://rangoon.ai/insights/archive-wayve-gaia-2-generative-world-model-march-2025/)
- [Meta paired concept segmentation with 3D reconstruction in its November 2025 SAM release](https://rangoon.ai/insights/archive-meta-sam-3-sam-3d-november-2025/)
- [Meta introduced SAM Audio with text, visual and temporal prompts for source separation](https://rangoon.ai/insights/archive-meta-sam-audio-december-2025/)

## More information

- [Documentation index](https://rangoon.ai/llms.txt)
- [AI access and policies](https://rangoon.ai/ai/)
- [Source repository](https://github.com/hypler-dev/rangoon)