Rent a GPU

Can the Apple M1 (16GB) run Llama 4 Scout 17B-16E?

Apple M1 (16GB) × Llama 4 Scout 17B-16E8K context
NoDoes not fit in memory
Memory at Q4_K_M68.3 GB of 10.7 GB
Weights 66.1 GBKV cache 1.6 GBRuntime 0.6 GB
  1. FP16
  2. Q8_0
  3. Q6_K
  4. Q5_K_M
  5. Q4_K_M
  6. Q3_K_M
  7. Q2_K

No. Even at Q4_K_M, Llama 4 Scout 17B-16E needs about 68.3 GB, but the Apple M1 (16GB) has 10.7 GB usable. You could offload some layers to system RAM, but it will be very slow.

Try other context lengths

Run it on a rented GPU

On a cloud GPU with enough memory, the same command you'd use locally works:

ollama run llama4:scout

At Q4_K_M it needs about 68.3 GB, so pick a GPU with at least 72 GB.

New to this? Step-by-step setup for the Apple M1 (16GB)

Run it in the cloud

Too big for the Apple M1 (16GB)? Run Llama 4 Scout 17B-16E on a rented GPU instead.

Rent a cloud GPU by the hour. No upfront cost, stop any time. Not sure whether to rent or buy? Compare the cost.

Rent a GPU on RunPodRent a GPU on Vast.ai

Memory needed by quantization

At 8,192 tokens of context. The black line marks the 10.7 GB the Apple M1 (16GB) can use. macOS lets the GPU use about 10.7 GB of the 16 GB unified memory by default.

QuantQualityMemoryFits?Speed
FP16Full quality220.2 GBToo big—
Q8_0Near-lossless118 GBToo big—
Q6_KExcellent91.6 GBToo big—
Q5_K_MVery good79.7 GBToo big—
Q4_K_MGood, most popular68.3 GBToo big—
Q3_K_MNoticeably degraded55.5 GBToo big—
Q2_KHeavily degraded43.1 GBToo big—

GPUs that run Llama 4 Scout 17B-16E comfortably

Models that run well on the Apple M1 (16GB)

Frequently asked questions

Can Apple M1 (16GB) run Llama 4 Scout 17B-16E?

No. Even at Q4_K_M, Llama 4 Scout 17B-16E needs about 68.3 GB, but the Apple M1 (16GB) has 10.7 GB usable. You could offload some layers to system RAM, but it will be very slow.

How much VRAM does Llama 4 Scout 17B-16E need?

About 68.3 GB at Q4_K_M with 8,192 tokens of context, and about 220.2 GB at full FP16 precision.

How fast is Llama 4 Scout 17B-16E on the Apple M1 (16GB)?

It does not fit in the Apple M1 (16GB)'s memory, so most layers would run from system RAM and generation would be very slow.