seoart/ARTART-v2

VISIONConcurrency Cost:1Model Size:5.1BQuant:BF16Ctx Length:32kTool Calling:SupportedPublished:Jun 23, 2026Architecture:Transformer Cold

ARTART-v2 is a 5.1 billion parameter GGUF model, fine-tuned and converted using Unsloth. This model is available in both text-only (F16) and multimodal (BF16-mmproj) variants, offering capabilities for both standard language tasks and vision-enabled applications. It is optimized for efficient deployment and use with tools like llama-cli, providing a versatile solution for various AI inference needs.

Loading preview...

ARTART-v2: Efficient GGUF Model

ARTART-v2 is a 5.1 billion parameter model, specifically fine-tuned and converted into the GGUF format using Unsloth. This optimization allows for faster training and efficient deployment.

Key Capabilities

  • Text-only Inference: The gemma-4-e2b-it.F16.gguf variant is designed for standard text-based language model tasks.
  • Multimodal Vision: The gemma-4-e2b-it.BF16-mmproj.gguf variant includes multimodal capabilities, supporting vision-enabled applications.
  • GGUF Format: Optimized for use with llama-cli and llama-mtmd-cli for straightforward execution.
  • Ollama Compatibility: Provides specific instructions for creating unified models compatible with Ollama, addressing its current limitation with separate mmproj files for vision models.

Good For

  • Developers seeking efficient, GGUF-formatted models for local inference.
  • Applications requiring both text-only and multimodal (vision) capabilities.
  • Users looking for models optimized with Unsloth for faster performance.

Finetunes
1 models