Skip to main content

Version 1.1

We are excited to announce the release of GenAI Studio v1.1.0, featuring significant enhancements to boost your GenAI development and deployment experience.

✨ What's New? ​

🔧 Revamped Management Interface & Optimized Architecture​

  • Entirely refactored GenAI Studio management interface
  • Enhanced architecture to better support Inference Apps such as Flux.1 Schnell (Text to Image) and ScrapeGraphAI Revamped Management Interface
  • Easier management and streamlined deployment of diverse GenAI applications

🔄 RAGOps Auto-Sync​

  • Automatic document synchronization from designated folders directly to your vector databases
  • Significantly improved RAG (Retrieval-Augmented Generation) workflow efficiency

📊 Real-time System Monitoring​

  • Integrated Grafana and Prometheus for real-time system performance tracking
  • Proactively detect and address potential issues before they escalate Real-time System Monitoring

🚀 Model Conversion & Inference Runtime​

  • Model conversion functionality has been added, along with the availability of the EdgeAI SDK for inference-side
  • downloads and deployment. This makes it easier to deploy models to edge devices.

🎯 Enhanced Fine-Tuning with LORA​

  • Integrated LORA (Low-Rank Adaptation) fine-tuning support via Unsloth
  • Compatible with inference models like DeepSeek for precise, customized adjustments Enhanced Fine-Tuning with LORA

⚙️ Upgrades & Maintenance ​

  • AnythingLLM updated to v1.7.5, providing the latest features and improved security.
  • Phison Firmware upgraded to NXUN202.00, enhancing hardware performance and stability.

Third-Party Updates​

GenAI Studio utilizes the following components:

  • node-exporter (1.8.2)
    Exposes desired host metrics to Prometheus.
  • dcgm-exporter (4.0.0-4.0.1-ubuntu22.04)
    Exposes host GPU metrics to Prometheus.
  • Prometheus (3.1.0)
    Collects desired metrics as a data source for Grafana.
  • Grafana (11.4.0)
    Serves as the resource monitoring dashboard.
  • Phison aiDAPTIVLink (NXUN202.00)
    Leverages middleware for model fine-tuning with full parameters.
  • Ollama (0.6.2)
    Acts as the inference server.
  • llama.cpp (full-cuda-b4897)
    Converts GGUF model file formats.
  • vsFTP (3.0.5)
    Provides model files for downloading.
  • Qdrant (1.12.4)
    Functions as the vector database.
  • Flowise (2.2.7-patch.1)
    Automates RAGOps functionality and workflows.
  • PostgreSQL (16.4)
    Serves as the relational database.
  • Unsloth (2025.3.18)
    Performs model fine-tuning with LoRA mode.