cd /news/robotics/scalevla-scaling-vision-language-act… · home › topics › robotics › article
[ARTICLE · art-171087] src=arxiv.org ↗ pub= topic=robotics verified=true sentiment=↑ positive

ScaleVLA: Scaling Vision-Language-Action Models to 552 Billion Parameters

A paper submitted to arXiv on 16 Nov 2026 introduces ScaleVLA, a Vision-Language-Action model that pairs a 552B-parameter DeepSeek-V4.1-Flash backbone with a trainable action DiT, and reports that frontier-scale VLM representations outperform smaller backbones on robot learning. ScaleVLA reached 87.8% with just 10 demonstrations versus 85.6% for a 2B backbone trained on 50, hit 45.1% success on LIBERO-PRO against at most 31.3% for smaller backbones, and raised real-world charger insertion from 4/10 to 9/10 successes while holding 216 ms median latency per action chunk. The authors argue frontier VLMs can serve as representational foundations for efficient, generalizable robot policies, not only as reasoning agents.

read2 min views5 publishedNov 17, 2026
ScaleVLA: Scaling Vision-Language-Action Models to 552 Billion Parameters
Image: source
  [Submitted on 16 Nov 2026]


[View PDF](http://arxiv.org/pdf/2611.21607v1)

[HTML (experimental)](https://arxiv.org/html/2611.21607v1)

Abstract:Frontier Vision-Language Models (VLMs) such as GPT-6 Astra have demonstrated remarkable capabilities in robot control, suggesting that broad pretraining provides knowledge valuable for physical action. Yet most Vision-Language-Action (VLA) policies still rely on much smaller VLM backbones, leaving this potential largely untapped. We ask whether frontier-scale representations can provide a stronger foundation for robot learning. We introduce ScaleVLA, which connects a 552B DeepSeek-V4.1-Flash backbone to a trainable action DiT, and systematically compare VLMs from 2B to 552B under a shared training setup. We evaluate three fundamental capabilities of robot foundation models. In sample efficiency, just 10 demonstrations outperform a 2B backbone trained with 50 (87.8% vs. 85.6%). In generalization, LIBERO-PRO success reaches 45.1%, compared with at most 31.3% for smaller backbones. In precise manipulation, real-world charger insertion improves from 4/10 to 9/10 successes, with gains concentrated in contact-sensitive alignment. Despite its scale, ScaleVLA achieves 216 ms median latency per action chunk. Our findings suggest that frontier VLMs can serve not only as reasoning agents, but also as powerful representational foundations for efficient and generalizable robot policies.

Additional Features

References & Citations

...

Bibliographic Explorer

(What is the Explorer?) Connected Papers

(What is Connected Papers?) Litmaps

(What is Litmaps?) scite Smart Citations

(What are Smart Citations?) alphaXiv

(What is alphaXiv?) CatalyzeX Code Finder for Papers

(What is CatalyzeX?) DagsHub

(What is DagsHub?) Gotit.pub

(What is GotitPub?) Hugging Face

(What is Huggingface?) ScienceCast

(What is ScienceCast?) Influence Flower

(What are Influence Flowers?) CORE Recommender

(What is CORE?) arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

── more in #robotics 4 stories · sorted by recency
── more on @scalevla 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
→ Live at https://your-agent.zahid.host ✓
Get free account → Pricing
from €0/mo · no card required
LIVE [news/scalevla-scaling-vis…] indexed:0 read:2min 2026-11-17 · —