Running Qwen 3.8 27B on Mac Studio Ultra

Qwen 3.8 27B has 27 billion parameters. Mac Studio Ultra offers 512 GB of Apple unified memory. This page puts the two side by side.

By QDNA · Published on 27 July 2026 · Updated on 15 September 2026

Short answer. Qwen 3.8 27B fits comfortably on Mac Studio Ultra. In FP16, its weights take about 62.1 GB of the 512 GB available, leaving 450 GB for the attention cache and concurrency.

How much memory does Qwen 3.8 27B need?

Weights take the parameter count multiplied by the format size, plus a 15% runtime margin for activations and buffers. In FP16 the calculation reads: 27 billion parameters × 2 bytes (FP16) = 54.0 GB of weights; × 1.15 runtime margin = 62.1 GB. The attention cache sits on top: it grows with context length and with the number of concurrent requests, so it is sized case by case.

Format	Weights in memory	On this machine	What the format costs
FP16	62.1 GB	fits	full precision, the quality reference
FP8	31.0 GB	fits	negligible loss on most tasks
NVFP4	15.5 GB	fits	Blackwell format, native FP4 compute
Q4	18.6 GB	fits	block quantisation, broadly supported

Model sheet: 27 billion parameters, dense architecture, 262,144-token context, Apache 2.0 licence. Dense model: every parameter computes on each token. The BF16 repository holds 27.8 billion elements (56 GB).

What is left to serve requests?

Qwen 3.8 27B fits comfortably on Mac Studio Ultra. In FP16, its weights take about 62.1 GB of the 512 GB available, leaving 450 GB for the attention cache and concurrency. The remaining margin decides how many concurrent requests and how much context the machine sustains. By QDNA convention a machine “fits” when the weights take at most 75% of its memory; beyond that it “barely fits” and forces a limit on context or on concurrency. The attention cache is not calculated on this page. These models’ architectures (compressed latent, linear or sparse attention, Mamba layers) share no per-token formula; it is measured on the machine, with the target context and concurrency [TO BE MEASURED].

Mac Studio Ultra: indicative price €6,599 to €20,679 incl. VAT, not contractual. Source: Apple Store France as reported by MacGeneration on 25 August 2026: €6,599 for the M5 Ultra with 96 GB, €20,679 with 256 GB and 16 TB; the 512 GB variant is not priced yet. The machine targets the TPE, PME segment.

Where else can Qwen 3.8 27B run?

See the Qwen 3.8 27B and Mac Studio Ultra fact sheets.

How much memory does Qwen 3.8 27B need?

In FP16, weights take about 62.1 GB including the runtime margin. The attention cache sits on top and depends on context.

Does Qwen 3.8 27B fit on Mac Studio Ultra?

Qwen 3.8 27B fits comfortably on Mac Studio Ultra. In FP16, its weights take about 62.1 GB of the 512 GB available, leaving 450 GB for the attention cache and concurrency.

Method and sources

Memory footprints are calculated, not measured: a reading on real hardware may differ depending on the engine and the exact weight format. Every figure below carries its source and the date of its last check.

Qwen 3.8 27B: 27 billion parameters, 262,144-token context, Apache 2.0 licence. Hugging Face repository (config.json, API, model card), checked on 2 September 2026. Dense model: every parameter computes on each token. The BF16 repository holds 27.8 billion elements (56 GB).
Mac Studio Ultra: 512 GB of memory; M5 Ultra, up to 512 GB unified memory, 1.2 TB/s, 480 W continuous; the 512 GB variant is announced for late October 2026. Manufacturer sheet checked on 2 September 2026.
Mac Studio Ultra price: €6,599 to €20,679 incl. VAT, indicative and not contractual. Source: Apple Store France as reported by MacGeneration on 25 August 2026: €6,599 for the M5 Ultra with 96 GB, €20,679 with 256 GB and 16 TB; the 512 GB variant is not priced yet.
Bytes per parameter: FP16 2 (16 bits per parameter, hence 2 bytes); FP8 1 (8 bits per parameter (E4M3), hence 1 byte, block scales not counted); NVFP4 0.5 (4 bits per parameter, hence 0.5 byte; 4-bit published weights weigh 0.54 to 0.56 byte per parameter with their scales (DeepSeek V4 Pro 865 GB for 1,599 billion, Kimi K3 1,561 GB for 2,780 billion)); Q4 0.6 (taken here as llama.cpp Q4_K_M, 0.6 byte per parameter including scales; Q4_K_S weighs 0.56, AWQ and GPTQ 0.55). Source: connaissance/faits.yaml, quantifications family, checked on 2026-08-31.
Runtime margin × 1.15: QDNA operating assumption, not measured: 15% above the weights for activations, buffers and fragmentation.
“Fits” threshold at 75% of memory: QDNA assumption, not measured, which keeps the remaining quarter for the attention cache and concurrency.