H Company Releases Holo4 Vision-Language Models for Computer Use

H Company has released Holo4, a pair of vision-language models (VLMs) designed specifically for Computer Use tasks. The models are distributed as GGUF files and are built on the Qwen3.8 dense architecture and Qwen3.5 mixture-of-experts architecture, respectively.

Model Variants

The release includes two distinct model sizes:

• Holo4-27B-GGUF: A dense model based on the Qwen3.8 architecture.
• Holo4-35B-A3B-GGUF: A mixture-of-experts model based on the Qwen3.5 architecture.

Integration and Capabilities

Both models are designed to work with the hai-agents harness. This integration allows the system to:

• Send screenshots and tool results to the model.
• Execute requested actions, including clicks, typing, code, and tool calls.

The models are available for download on Hugging Face under the Hcompany organization.