By memory size
| Memory | Run well | Biggest that runs well | gpt-oss-20b |
|---|---|---|---|
| 16 GB153 GB/s | 20 | Gemma 4 12B Unified | 43–77 tok/s |
| 24 GB170 GB/s | 24 | Gemma 4 26B A4B | 40–72 tok/s |
| 32 GB170 GB/s | 29 | Qwen3.6 35B A3B | 40–72 tok/s |
Models: Mac mini (M6 or M5 Pro). Not every memory size is offered with every model; Apple’s Tech Specs list what each one comes with.
Questions
- Is the Mac mini with M6 good for local LLMs?
- With 32 GB it runs 29 of 73 open models well, up to Qwen3.6 35B A3B, and writes about 40–72 tokens a second with gpt-oss-20b. The M6 reads memory at 170 GB/s, which sets how fast it writes.
- How much memory should a Mac mini with M6 have for AI?
- With 16 GB it runs 20 open models well; with 32 GB, 29, up to Qwen3.6 35B A3B. Memory decides which models fit, and it can’t be upgraded later.
Other Macs
The Mac mini with other chips:
Head to head:
The M6 for local AI · The best Mac for local LLMs · Every Mac
How it’s worked out
Memory: the model’s parameters at 8, 6, 4, 3 or 2 bits per weight plus MLX’s group scales (half a bit per weight), or its native format, with room for the context on top. By default macOS lets the GPU use about 75% of unified memory. Speed: memory bandwidth divided by the bytes read per token is the ceiling for one conversation; in llama.cpp’s public Apple Silicon benchmarks real runs reach about 45% to 80% of it, which is the range shown. These are estimates, not measurements: measure your own Mac on MLX Bench. Model figures from each official model card, chip figures from Apple’s Tech Specs. Checked 2026-10-19. Same data as mlx.app/compare.