selfhost.directory

The index of self-hostable projects — ranked, categorized and served with live updates.
Own your data and stop giving it to big tech.

inclusionAI logo

Can the Apple M4 (16 GB) run ZwZ2-9B?

ZwZ2-9B·9.4B·Apple M4 (16 GB)·16 GB unified memory

Saved in this browser, used on every model page.
Yes

Yes. The Apple M4 (16 GB) runs ZwZ2-9B fully in unified memory at Q6_K (about 8.7 GB of 11 GB), at roughly 8.5 tokens/s.

best quant
Q6_K
needs
8.7 GB
usable
11 GB
est. speed
~8.5 tok/s

Estimates for an 8K context; offload assumes 32 GB of system RAM.Try other contexts on the ZwZ2-9B page →

Every quantisation of ZwZ2-9B on the Apple M4 (16 GB)

QuantQualityMemoryFitSpeed
Q2_K Heavy loss4.5 GBfits~21 tok/s
Q3_K_M Noticeable loss6.0 GBfits~14 tok/s
Q4_K_M Small loss7.0 GBfits~12 tok/s
Q5_K_M Near-lossless7.8 GBfits~9.9 tok/s
Q6_K pickNear-lossless8.7 GBfits~8.5 tok/s
Q8_0 Lossless11 GBtoo big—
FP16 / BF16 Lossless19 GBtoo big—

Speed is a bandwidth-based estimate for generating text (120 GB/s); prompt processing and long contexts are slower.

Measured speed on the Apple M4 (16 GB)

No community reports for this pair yet.

Report your own speed →

Run ZwZ2-9B on the Apple M4 (16 GB)

Using Q6_K, the best fit for Apple M4 (16 GB).

llama.cpp · OpenAI-compatible server with a web UIdocs
llama-server -hf prithivMLmods/inclusionAI-ZwZ2-9B-GGUF:Q6_K

brew install llama.cpp ships with Metal. Downloads the GGUF on first run.

Transformers · Ask about an imagedocs
from transformers import pipeline

pipe = pipeline("image-text-to-text", model="inclusionAI/ZwZ2-9B", dtype="auto", device_map="auto")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/bee.jpg"},
            {"type": "text", "text": "Describe this image."},
        ],
    }
]
out = pipe(text=messages, max_new_tokens=256)
print(out[0]["generated_text"][-1]["content"])

Vision models that fit the Apple M4 (16 GB) comfortably

All that fit →
ModelParamsBest quantMemorySpeedLicence
9.7BQ6_K≈8.8 GBFits in memory~8.3 tok/sCommercial OK
8.8BQ8_0≈10 GBFits in memory~7.1 tok/sCommercial OK
9.4BQ6_K≈9.1 GBFits in memory~8.5 tok/sCommercial OK
8.3BQ8_0≈8.9 GBFits in memory~7.5 tok/sCommercial OK
Cloudflare logoclef-flashCloudflare
9.4BQ6_K≈9.2 GBFits in memory~8.5 tok/sCommercial OK
10.7BQ5_K_M≈9.2 GBFits in memory~8.7 tok/sConditions

Other GPUs that run ZwZ2-9B

Frequently asked questions

Can the Apple M4 (16 GB) run ZwZ2-9B?
Yes. The Apple M4 (16 GB) runs ZwZ2-9B fully in unified memory at Q6_K (about 8.7 GB of 11 GB), at roughly 8.5 tokens/s.
How fast does ZwZ2-9B run on the Apple M4 (16 GB)?
An estimate from memory bandwidth gives about 8.5 tokens/s at Q6_K.
Which quantisation of ZwZ2-9B should I use on the Apple M4 (16 GB)?
Q6_K: the largest one that fits. Expected quality: near-lossless compared with the full-precision weights.

Memory and speed are estimates from the model's size and the card's memory bandwidth; real numbers depend on the runtime, drivers and context length. Community figures are medians of user reports.