All in one lite version! 8B parameter instead of Flux's 12b.

This is an all-in-one (AIO) model, embedded are files (decoders) t5xxl_fp8_e4m3fn_scaled.safetensors, clip_l_large.safetensors and flux_vae.safetensors.

One of the first open-source 8B image models capable of generating coherent images without needing more external loading files. Great for GPUs with 8 to 12 GB VRAM. FP8 quantized from fp16.