microsoft/GELab-Zero-4B-preview-Sico-Evolution
microsoft/GELab-Zero-4B-preview-Sico-Evolution is a 4 billion parameter GUI agent, fine-tuned with LoRA from the GELab-Zero-4B-preview base model on Microsoft Edge and Copilot UI trajectories. Developed by Microsoft as part of the Sico platform, it utilizes a general-purpose GUI model evolution pipeline to iteratively improve task success rates. This model excels at GUI automation, achieving a dominant 82.9% Task Success Rate, outperforming several closed-source and open-source models in its domain.
Loading preview...
GELab-Zero-4B-preview-Sico-Evolution: An Evolved GUI Agent
This model, developed by Microsoft as part of the Sico platform, is a 4 billion parameter GUI agent. It is fine-tuned (LoRA) from the GELab-Zero-4B-preview base model specifically on Microsoft Edge and Copilot UI trajectories. The core innovation lies in its general-purpose GUI model evolution pipeline, an iterative mechanism designed to continuously enhance an agent's real task success rate across various GUI applications.
Key Capabilities & Performance
- Significant Performance Improvement: Sico-Evolution demonstrates a remarkable 82.9% Task Success Rate, representing a +43.1% absolute surge over the base model's 39.8% baseline.
- Outperforms Leading Models: It surpasses several top proprietary models, including gpt-5.4 (79.7%), Claude-Opus-4.6 (81.3%), and claude-opus-4.7 (82.1%).
- Dominates Open-Source Competitors: The model also significantly outperforms other leading open-source models like kimi-k2.6 (62.6%) and UI-Venus-1.5-30B (61.0%).
Use Cases
This model is particularly well-suited for:
- Automating GUI interactions within applications like Microsoft Edge and Copilot.
- Developing Digital Workers that can co-evolve with human operators through real-world tasks.
- Research into agentic evolution and co-evolving human-AI systems, as detailed in Microsoft's survey publication.