Skip to content

Can Mac M2 Max 96G run Arete 27B?

Comfortable: GGUF Q8_0 needs 30.2 GB at 4K, batch 1; headroom 41.8 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_M14.0ComfortableVerified file
GGUF Q4_K_M18.3ComfortableVerified file
GGUF Q5_K_M21.9ComfortableVerified file
GGUF Q8_030.2ComfortableVerified file

Context capacity · GGUF Q8_0

ContextTotal GBVerdict
4K30.2Comfortable
16K31.1Comfortable
32K32.1Comfortable
128K38.8Comfortable

Generation speed

GGUF Q8_0: 8–11 tok/s · Derived range

Other model / GPU pairs