AP Apple Prosumer Unified memory

Apple M1 Ultra 128GB

Runs downloadable LLMs up to ~195B at Q4, ~121.4-146.4 tok/s on an 8B model.

Other configs: 64GB
Apple M1 Ultra 128GB
128GB
Unified memory
800
GB/s bandwidth
195B
Max model (Q4)
$4,800
MSRP

What you can run

At 8k context, FP16 KV. Full-speed tok/s holds up to "max ctx"; beyond it the context spills to SSD swap and drops to the degraded figure.

Model sizeFP16Q8Q4
3B 90.7-109.3 t/s
≤1643k ctx · then ~1.1-1.4 (slow)
181.3-218.7 t/s
≤1689k ctx · then ~2.3-2.7 (slow)
323.8-390.5 t/s
≤1709k ctx · then ~4-4.9 (slow)
7-8B 34-41 t/s
≤745k ctx · then ~0.4-0.5 (slow)
68-82 t/s
≤806k ctx · then ~0.9-1 (slow)
121.4-146.4 t/s
≤833k ctx · then ~1.5-1.8 (slow)
13B 20.9-25.2 t/s
≤535k ctx · then ~0.3-0.3 (slow)
41.8-50.5 t/s
≤615k ctx · then ~0.5-0.6 (slow)
74.7-90.1 t/s
≤650k ctx · then ~0.9-1.1 (slow)
32B 8.5-10.3 t/s
≤190k ctx · then ~0.1-0.1 (slow)
17-20.5 t/s
≤312k ctx · then ~0.2-0.3 (slow)
30.4-36.6 t/s
≤365k ctx · then ~0.4-0.5 (slow)
70B won't run 7.8-9.4 t/s
≤133k ctx · then ~0.1-0.1 (slow)
13.9-16.7 t/s
≤227k ctx · then ~0.2-0.2 (slow)
120B won't run won't run 8.1-9.8 t/s
≤142k ctx · then ~0.1-0.1 (slow)
400B won't run won't run won't run

Estimate: tok/s ≈ bandwidth ÷ (params × bytes/param) × efficiency (Apple). Cyan = fits ~115.2 GB usable. Beyond "ctx", KV spills to SSD swap (rough degraded speed shown).

Run popular models on the Apple M1 Ultra 128GB

Videos about the Apple M1 Ultra 128GB

Is M1 Ultra enough for MACHINE LEARNING? vs RTX 3080ti

NO other PC can match it for $4,000? - M1 Ultra Mac Studio review

More hardware like the Apple M1 Ultra 128GB

About the Apple M1 Ultra 128GB

Apple M1 Ultra 128GB is a unified-memory system from Apple on Apple M1, 2022. For local genAI the numbers that matter are its 128 GB (how big a model + context fits) and 800 GB/s bandwidth (generation speed).

Frequently asked questions

What size LLM can the Apple M1 Ultra 128GB run?

With ~115.2 GB usable it runs up to ~195B at Q4 (~55.5B at FP16) at 8k context. Bigger spills into SSD swap and slows sharply.

How many tokens/sec does the Apple M1 Ultra 128GB generate?

Generation is bandwidth-bound. At 800 GB/s an 8B model at Q4 runs ~121.4-146.4 tok/s (estimate).

Why does long context slow it down?

The KV cache for the context must also fit fast memory; once weights+KV exceed 115.2 GB it spills to SSD swap and throughput collapses. Each row shows the max usable context.