Skip to content

Can Mac M2 Ultra 192G run Nanbeige4.3 33B?

Comfortable: GGUF Q8_0 needs 37.7 GB at 4K, batch 1; headroom 106.3 GB.

Text input, FP16 KV cache, one card. Unified-memory capacity uses the GPU guide’s usable limit; negative headroom may require host offload. Speed source settings may differ from this 4K estimate.

Measured runs

No measured run for this pair. Sizing is an estimate, not a runtime test.

Quant memory at 4K

QuantTotal GBVerdictWeight sizing
GGUF Q3_K_M17.6ComfortableVerified file
GGUF Q4_K_M22.6ComfortableVerified file
GGUF Q8_037.7ComfortableVerified file

Context capacity · GGUF Q8_0

ContextTotal GBVerdict
4K37.7Comfortable
16K40.4Comfortable
32K43.9Comfortable
128K65.4Comfortable

Generation speed

GGUF Q8_0: 24 tok/s · Bandwidth ceiling (not measured)

Other model / GPU pairs