Skip to content

Can Mac M1 Max 64G run Recursive Flint 40B-A2.5B?

Comfortable: GGUF Q4_K_M needs 26.7 GB at 4K, batch 1; headroom 21.3 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_M20.6ComfortableVerified file
GGUF Q4_K_M26.7ComfortableVerified file
GGUF Q8_044.5TightVerified file

Context capacity · GGUF Q4_K_M

ContextTotal GBVerdict
4K26.7Comfortable
16K27.0Comfortable
32K27.5Comfortable
128K30.4Comfortable

Generation speed

GGUF Q4_K_M: 265 tok/s · Bandwidth ceiling (not measured)

Other model / GPU pairs