Newsroom
Holo2 235B
Holo2-235B-A22B Preview is H Company's largest UI localization model yet, setting new state-of-the-art results on ScreenSpot-Pro and OSWorld G.
February 3, 2026 · 1 min read
New Holo2 model takes the lead in UI Localization
Two months since releasing the first Holo2 models, we're back with our largest UI localization model yet: Holo2-235B-A22B Preview. This model achieves a new state-of-the-art of 78.5% on ScreenSpot-Pro and 79.0% on OSWorld G.
Holo2-235B-A22B records the highest average score across six UI localization benchmarks.
Holo2-235B-A22B Preview is a research release focused on UI element localization. It's available today on Hugging Face.
Agentic Localization
High-resolution 4K interfaces are challenging for localization models. Small UI elements can be difficult to pinpoint on a large display. With agentic localization, Holo2 can iteratively refine its predictions, improving accuracy with each step and unlocking 10-20% relative gains across all Holo2 model sizes.
Holo2-235B-A22B Preview reaches 70.6% accuracy on ScreenSpot-Pro in a single step. In agent mode, it achieves 78.5% within 3 steps, setting a new state-of-the-art on the most challenging GUI grounding benchmark.
Holo2-235B-A22B reaches 78.5% on ScreenSpot-Pro within three agentic localization steps.
Trained with SkyPilot
Training Holo2 models at scale requires coordinating workloads across multiple cloud providers. H Company uses SkyPilot as a unified interface for launching training jobs on our clusters with Kubernetes (k8s). By abstracting away infrastructure complexity, SkyPilot lets researchers focus on model development instead of managing k8s manifests or maintaining separate deployment scripts.
