Lumina2-q4_0-4Bit

Accuracy vs. Size Tradeoff

Every variant is benchmarked against the FP16 baseline. Choose the right balance of accuracy and footprint for your workflow.

Usage Code Snippets

Which variant for your hardware?

Choose the right quantization path for your device and performance target.

About This Model

Lumina Image 2.0 is an open-weight text-to-image foundation model developed by the Alpha-VLLM team. Built on the Unified Next-DiT architecture, the model treats text and image tokens as a unified sequence, enabling strong cross-modal understanding and prompt adherence. Despite containing only approximately 2.6 billion parameters in its image generation backbone, Lumina 2.0 achieves competitive image quality while requiring substantially fewer resources than larger models such as FLUX, HiDream, and Qwen-Image. The model utilizes a Gemma 2B text encoder, a Flow Matching diffusion framework, and a Flux-compatible VAE. The GGUF quantized version enables efficient local deployment through ComfyUI-GGUF and Diffusers while significantly reducing memory requirements. Lumina 2.0 is particularly attractive for users seeking a balance between image quality, speed, hardware efficiency, and commercial-friendly Apache 2.0 licensing.