Hcompany/Holo-3.1-35B-A3B
Holo-3.1-35B-A3B is a 35 billion parameter Vision-Language Model (VLM) developed by H Company, built upon the Qwen 3.5 family. This model is specifically designed for computer use agents, expanding support to web, desktop, and mobile environments. It features native function-calling capabilities and offers optimized quantized checkpoints for local deployment. Holo-3.1-35B-A3B excels in UI grounding, mobile automation, and enterprise workflows, providing a cost-efficient solution for agent-based applications.
Loading preview...
Holo3.1: Vision-Language Models for Computer Use Agents
Holo3.1 is H Company's latest family of Vision-Language Models (VLMs), extending the capabilities of Holo3. These models are engineered for computer use agents, supporting automation across web, desktop, and mobile environments. A key differentiator is the introduction of native function-calling support, facilitating seamless integration with various agent frameworks. The Holo3.1 family, built on the Qwen 3.5 base models, ranges from 0.8B to 35B-A3B parameters, with the Holo-3.1-35B-A3B variant offering optimized quantized checkpoints (BF16, FP8, NVFP4, Q4 GGUF) for efficient local deployment.
Key Capabilities
- Multi-Environment Automation: Supports agent operations across web, desktop, and mobile interfaces.
- Native Function-Calling: Designed for direct integration with agent frameworks.
- Optimized for Local Deployment: Available with various quantizations for cost-effective, on-device inference.
- Strong Performance: Delivers robust results across computer use, UI grounding, mobile automation, and enterprise workflow benchmarks.
Good For
- Developing fast and local computer use agents.
- Applications requiring UI grounding and mobile automation.
- Enterprise workflows where cost-efficiency and deployment flexibility are critical.
- Integrating vision-language capabilities into agent frameworks with native function-calling.