Hcompany/Holo-3.1-35B-A3B
TEXT GENERATIONConcurrency Cost:3Model Size:35.1BQuant:FP8Ctx Length:32kTool Calling:SupportedPublished:May 29, 2026License:apache-2.0Architecture:Transformer0.1K Open Weights Cold
Holo-3.1-35B-A3B is a 35.1 billion parameter Vision-Language Model (VLM) developed by H Company, built upon the Qwen 3.5 family. Optimized for computer use agents, it supports web, desktop, and mobile automation with native function-calling. This model is designed for local deployment through various quantized checkpoints, offering a balance of performance and cost efficiency across diverse agent workflows.
Loading preview...
Holo3.1-35B-A3B: Vision-Language Model for Computer Use Agents
Holo3.1-35B-A3B is a 35.1 billion parameter Vision-Language Model (VLM) from H Company, part of the Holo3.1 family built on the Qwen 3.5 architecture. This model is specifically designed for computer use agents, expanding capabilities beyond traditional browser and desktop automation to include mobile environments.
Key Capabilities & Features
- Multi-Environment Automation: Supports agent operations across web, desktop, and mobile platforms.
- Native Function-Calling: Integrates seamlessly with agent frameworks through built-in function-calling support.
- Local Deployment Optimization: Available in various quantized checkpoints (BF16, FP8, NVFP4, Q4 GGUF) for efficient local deployment.
- Performance & Cost Efficiency: Demonstrates strong performance across computer use, mobile automation, enterprise workflows, and UI grounding benchmarks, establishing a favorable Pareto frontier for inference cost.
Ideal Use Cases
- Developing intelligent agents for automating tasks on computers and mobile devices.
- Applications requiring UI grounding and interaction across different operating environments.
- Enterprise workflows needing robust and cost-effective local VLM deployment.
- Scenarios where a balance of high performance and deployment flexibility is crucial.