Quick Answer

A local AI and Stable Diffusion rig in Mossel Bay is decided by GPU VRAM. The RTX 4060 Ti 16 GB (~R10,500) runs SDXL, Flux and ComfyUI without out-of-memory errors; an RTX 5070 Ti 16 GB or RTX 4090 24 GB is the step up for speed and video models. Pair the GPU with a Ryzen 5 7600 and 32 GB DDR5. VRAM, not CPU speed, sets the model size you can run.

VRAM tiers and SD models

Stable Diffusion loads the model into GPU memory, so VRAM is the wall:

  • 8 GB: SD 1.5 only
  • 12 GB: SDXL comfortable, Flux tight
  • 16 GB (RTX 4060 Ti 16 GB / RTX 5070 Ti): SDXL, Flux.1, ComfyUI batches
  • 24 GB (RTX 4090): large batches and video models

For most local-AI hobbyists, 16 GB runs SD 1.5, SDXL and Flux comfortably.

The supporting build and Garden Route delivery

  • CPU: Ryzen 5 7600 (~R4,200) - feeds the GPU
  • RAM: 32 GB DDR5-6000 (~R2,000)
  • Storage: 1 TB NVMe (~R1,200) - checkpoints are 2-7 GB each
  • PSU: 650-750 W 80+ Gold for stable long renders

Mossel Bay is on the Garden Route; coastal humidity means good ventilation. Power-test on arrival. For current delivery times and costs, see our delivery page.

FAQ

Which Stable Diffusion versions can I run?

A 16 GB card runs SD 1.5, SDXL and Flux.1 in ComfyUI or Automatic1111 with reasonable batch sizes. SD 1.5 is fastest; SDXL and Flux are heavier but comfortable, where 8 GB cards run out of memory.

Should I use ComfyUI or Automatic1111?

ComfyUI uses VRAM more efficiently and handles Flux and SDXL batches better, making it the better choice on a 16 GB card. Automatic1111 is simpler for beginners but more memory-hungry on large models.

Does Evetech deliver to Mossel Bay?

Yes, Evetech ships to Mossel Bay on the Garden Route. Power-test the GPU and confirm it's detected on arrival.

TIP

with fp16 and tiled VAE enabled on a 16 GB card - it keeps VRAM use low enough to generate larger SDXL and Flux images without the out-of-memory errors that stop 8 GB cards.