LLM on Mac / Model

Wan2.2 TI2V-5B

app-specific / offload

Starting memory estimate

64 GB

For trying short tasks one at a time

Experimental on Mac. The official 24GB VRAM setup targets CUDA GPUs, not Macs. Our 36GB machine recorded a 45.72GB peak, so 64GB+ is a cautious planning budget, not a speed guarantee.

Tasks to try

An experimental candidate for short motion and composition studies. Generate a few-second clip from an image and select usable parts.

Keep under review

Not recommended when you need a finished video quickly. Expect a Mac-compatible implementation, ample memory and long waits.

Getting started

Memory figures are editorial budgets including room for the OS and apps, not verified minimums. Text models assume 4-bit weights, 4K–8K input tokens and one model at a time. Long documents, images and concurrent apps need more. A3B does not mean the entire model fits in 3GB.

Assumes an Apple Silicon Mac and a compatible app: LM Studio or similar for text, Whisper apps for audio, and model-specific image/video tools. RAM alone does not determine speed. Start with a short input on your current Mac.

Measurements for this model

2026-09-24 JST · Apple M5 Max · 36 GB

Processing time852.7 s
Peak memory45.72 GB

Timing is per approximately 3.4-second clip; memory is the backend-reported peak. The peak exceeds the 36GB machine’s RAM, so this is not evidence of comfortable operation at 36GB.

Test conditions
{
  "width": 1280,
  "height": 704,
  "frames": 81,
  "fps": 24,
  "seed": 1,
  "mlx_cache_limit_gb": 2,
  "seconds": 3.4,
  "quantization": "—",
  "runtime": "—"
}

Generated examples

Generated example 1
Generated example 2
Generated example 3