ZERO-POINT-AI/Miss-MARTHA-hot-POCKET-edition-2B-OMNI

VISIONConcurrency Cost:1Model Size:2.3BQuant:BF16Ctx Length:32kTool Calling:SupportedPublished:Apr 6, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Cold

The ZERO-POINT-AI/Miss-MARTHA-hot-POCKET-edition-2B-OMNI is a 2.3 billion parameter multimodal AI model developed by Zero Point Intelligence Ltd, fine-tuned from Qwen/Qwen3.5-2B. This model supports image-text-to-text generation, allowing it to see and describe images, answer questions, and perform tasks like code generation and translation. Optimized for local deployment, it runs efficiently on personal computers with varying VRAM configurations, making it suitable for on-device AI companionship and entertainment.

Loading preview...

MARTHA-2B-OMNI: A Multimodal AI for Local Deployment

MARTHA-2B-OMNI, developed by Zero Point Intelligence Ltd, is a 2.3 billion parameter vision-language AI model based on Qwen/Qwen3.5-2B. It is designed for image-text-to-text generation, enabling it to interpret images, answer diverse questions, and handle tasks such as code writing, mathematical problem-solving, and language translation. A key differentiator is its optimization for local execution, supporting various GGUF quantization levels (Q8_0, Q6_K, Q5_K_M, Q4_K_M, Q3_0, Q2_0) to accommodate different hardware specifications, from high-VRAM GPUs to CPUs.

Key Capabilities

  • Multimodal Understanding: Processes both text and image inputs, allowing it to describe images and answer questions about visual content.
  • Local Operation: Designed to run on personal computers using tools like Ollama, eliminating the need for cloud services, API keys, or subscriptions.
  • Versatile Task Performance: Capable of general-purpose conversational AI, code generation, mathematical computations, and translation.
  • Privacy-Focused: Emphasizes on-device processing with no data harvesting, offering a private AI companion experience.
  • Robust Safety Measures: Features a strict, non-negotiable refusal mechanism for any content involving minors, prioritizing child safety.

Good For

  • On-device AI applications: Ideal for users who want to run a capable multimodal AI directly on their laptop or desktop.
  • Adult companionship and entertainment: Provides an orientation-neutral conversational AI for private, adult-oriented interactions.
  • Developers and hobbyists: Offers a flexible model with various quantization options for experimentation and integration into local projects.
  • Use cases requiring image understanding: Suitable for tasks where the AI needs to interpret and respond to visual information without cloud dependency.

Finetunes
1 models