Meet the moeinGTS Family: From Edge-Ready LLMs to Fine-Tuned Vision Adapters

The moeinGTS team, led by Ali and Arshiya Sohrevardi, has released an open-source family of AI models designed for local deployment, including lightweight LLMs (1.5b and 3b), a flagship model (GTS-1), a vision adapter (MoeinGTS-kamal-1), and a security adapter (moeinGTS-paspan). The models support GGUF quantization and integrate with Ollama, Hugging Face Diffusers, and PyTorch/TensorFlow, enabling collaborative multi-model agentic workflows.

🚀 Introducing the moeinGTS AI Family: Lightweight, Modular & Open-Source Building AI solutions often requires striking a balance between local efficiency, specialized performance, and seamless system integration. The moeinGTS ecosystem was designed to tackle these challenges by offering a suite of tailored open-source models—ranging from high-speed local LLMs to domain-specific vision and security adapters. Here is a full breakdown of the moeinGTS model lineup: 🤖 The Model Lineup - moeinGTS 1.5b — The Ultra-Lightweight Core - Role: Fast, low-latency reasoning & local orchestration. - Best For: Edge devices, quick zero-shot responses, microcontrollers, and resource-constrained environments. - moeinGTS 3b — The Balanced Workhorse - Role: High-efficiency general reasoning and structured data generation. - Best For: Interactive local applications, lightweight agentic pipelines, and local API backends. - GTS-1 — The Flagship Intelligence - Role: Advanced logic, complex prompt adherence, and multi-step execution. - Best For: Core system backends, deep code synthesis, and multi-agent coordination. - MoeinGTS-kamal-1 — The Visual & Artistic Adapter - Role: Specialized Diffusers/LoRA fine-tune for high-detail atmospheric and monochrome image generation. - Best For: Style-consistent visual generation, portraiture, and aesthetic UI graphics. - moeinGTS-paspan — The Security & Guardrail Sentinel - Role: Input validation, prompt injection defense, and output safety filtering. - Best For: Securing local agent workflows, auditing inputs, and ethical hacking guardrails. 🛠️ Architecture & Deployment The entire moeinGTS suite is built with local accessibility in mind: - Quantization Support: Optimized for GGUF formats Q4 K M , Q8 0 for low-VRAM environments. - Framework Integration: Plug-and-play compatible with Ollama , Hugging Face Diffusers , and PyTorch/TensorFlow pipelines. - Agentic Multi-Model Setup: Designed to run collaboratively—where paspan guards, GTS-1 plans, and kamal-1 generates visual assets. 🔗 Explore & Connect 🔗 Explore & Connect - Hugging Face Models: - Ollama Hub: arshiyasohrevardimoein/moeinGTS https://ollama.com/arshiyasohrevardimoein/moeinGTS - Developed by: Ali & Arshiya Sohrevardi moeinGTS Team AI OpenSource MachineLearning HuggingFace Python LocalAI