Rent a GPU

Can the Apple M4 Pro (48GB) run Mixtral 8x7B?

Apple M4 Pro (48GB) × Mixtral 8x7B8K context
YesRuns fully on the GPU
Memory at Q5_K_M34.9 GB of 36 GB
Weights 33.2 GBKV cache 1.1 GBRuntime 0.6 GB
QuantQ5_K_MVery good
Speed~18tokens per second
Max context15Ktokens that fit at Q5_K_M
  1. FP16
  2. Q8_0
  3. Q6_K
  4. Q5_K_M
  5. Q4_K_M
  6. Q3_K_M
  7. Q2_K

Yes. The Apple M4 Pro (48GB) can run Mixtral 8x7B fully in unified memory at Q5_K_M (very good), using about 34.9 GB of the 36 GB available, at roughly 18 tokens/second.

Try other context lengths

Run it with Ollama

With Ollama installed, one command downloads and starts Mixtral 8x7B:

ollama run mixtral:8x7b

Ollama's default download is Q4_K_M (about 30 GB here). You have room for Q5_K_M, which is very good: pick that GGUF in LM Studio or from Hugging Face.

New to this? Step-by-step setup for the Apple M4 Pro (48GB)

Run it in the cloud

No Apple M4 Pro (48GB)? Run Mixtral 8x7B in the cloud.

Rent a cloud GPU by the hour. No upfront cost, stop any time.

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 36 GB the Apple M4 Pro (48GB) can use. macOS lets the GPU use about 36 GB of the 48 GB unified memory by default.

QuantQualityMemoryFits?Speed
FP16Full quality95.1 GBToo big—
Q8_0Near-lossless51.3 GBToo big—
Q6_KExcellent40 GBToo big—
Q5_K_MVery good34.9 GBFits~18 tok/s
Q4_K_MGood, most popular30 GBFits~21 tok/s
Q3_K_MNoticeably degraded24.5 GBFits~26 tok/s
Q2_KHeavily degraded19.2 GBFits~34 tok/s

Measured speed from real Apple M4 Pro (48GB) owners

Tokens per second that people measured on their own Apple M4 Pro (48GB), next to our estimate.

No one has measured Mixtral 8x7B on the Apple M4 Pro (48GB) yet. The speeds above are estimates. Run the benchmark and yours becomes the first real number on this page.

Have an Apple M4 Pro (48GB)? Add your real speed

One command times Mixtral 8x7B in your own Ollama, shows you exactly what it would send, and asks before sending anything. No sign-up, no personal data.

macOS and Linux

curl -fsSL https://gpufit.dev/bench.sh | sh -s -- mixtral:8x7b

Windows (PowerShell)

$env:GPUFIT_MODEL="mixtral:8x7b"; irm https://gpufit.dev/bench.ps1 | iex

How the benchmark works

GPUs that run Mixtral 8x7B comfortably

Models that run well on the Apple M4 Pro (48GB)

Frequently asked questions

Can Apple M4 Pro (48GB) run Mixtral 8x7B?

Yes. The Apple M4 Pro (48GB) can run Mixtral 8x7B fully in unified memory at Q5_K_M (very good), using about 34.9 GB of the 36 GB available, at roughly 18 tokens/second.

How much VRAM does Mixtral 8x7B need?

About 30 GB at Q4_K_M with 8,192 tokens of context, and about 95.1 GB at full FP16 precision.

How fast is Mixtral 8x7B on the Apple M4 Pro (48GB)?

Roughly 18 tokens per second at Q5_K_M, estimated from the Apple M4 Pro (48GB)'s 273 GB/s memory bandwidth.