/research
/research

Research

Data research from HUD: how we measure the quality of AI training data, and the quality engine behind the HUD platform and DataVendor.

FeaturedRobotics

Evaluating Robot Models in the Cloud

HUD is partnering with Dreamscale Labs so robot developers can benchmark and evaluate frontier physical AI models in the cloud: 128 parallel LIBERO rollouts, world action models on RoboLab, and real-time π0.5 control on a bimanual YAM.

October 16, 2026 · HUD Team, Dreamscale Team

DateTitleFieldAuthors
  1. Oct 16, 2026Evaluating Robot Models in the CloudRoboticsHUD Team, Dreamscale Team
  2. Sep 28, 2026Can you trust your browser eval?Browser agentsHUD Team, Browserbase Team
  3. Sep 22, 2026What is a good task?EvalsReinis Salaks, Ilya Kuhta, Nancy Lau
  4. Jul 31, 2026Assemble Bench: Benchmarking Robot Models on Contact-Rich AssemblyRoboticsLukass Kellijs, The HUD Team
  5. Jan 20, 2026Building an RL Environment to Train Agents for Production DebuggingEngineeringThe HUD Team
  6. Oct 1, 2025Evaluating Agents on Financial Analyst Workflows (SheetBench)EnterpriseThe HUD Team, Sepal AI

Build environments that frontier labs trust

Create your first environment in minutes. No credit card required.

Anything you can simulate and grade, you can improve.

GitHubXDiscord
Product
  • Evaluations(opens in new tab)
  • Environments
  • DataVendor(opens in new tab)
Resources
  • Documentation(opens in new tab)
  • Blog
  • Research
  • Guides
Company
  • Careers30
  • Contact
© 2026 Human Union Data, Inc.
PrivacyTermsTrust(opens in new tab)