According to Vitalik Buterin, the FLUX development inference process takes approximately 5 minutes due to the limitations of his 4070 GPU, which has only 8 GB of VRAM. This necessitates the use of enable_sequential_cpu_offload(), resulting in slower performance. He mentioned that using a 4-bit quantization might fit but hasn't been tested yet. Additionally, manual edits take around 10 minutes, and a second round of inpainting takes about 1 minute.