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.

XHToken logo

Can the Apple M4 (16 GB) run Spark-X2.6-1.7B?

Spark-X2.6-1.7B·1.7B·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 Spark-X2.6-1.7B fully in unified memory at FP16 / BF16 (about 4.3 GB of 11 GB), at roughly 19 tokens/s.

best quant
FP16 / BF16
needs
4.3 GB
usable
11 GB
est. speed
~19 tok/s

Estimates for an 8K context; offload assumes 32 GB of system RAM.Try other contexts on the Spark-X2.6-1.7B page →

Every quantisation of Spark-X2.6-1.7B on the Apple M4 (16 GB)

QuantQualityMemoryFitSpeed
Q2_K Heavy loss1.5 GBfits~118 tok/s
Q3_K_M Noticeable loss1.8 GBfits~79 tok/s
Q4_K_M Small loss2.7 GBfits~64 tok/s
Q5_K_M Near-lossless2.1 GBfits~54 tok/s
Q6_K Near-lossless2.3 GBfits~47 tok/s
Q8_0 Lossless2.7 GBfits~36 tok/s
FP16 / BF16 pickLossless4.3 GBfits~19 tok/s

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 Spark-X2.6-1.7B on the Apple M4 (16 GB)

Using Q8_0, the best fit for Apple M4 (16 GB).

MLX · Apple Silicon with mlx-lmdocs
pip install mlx-lm
mlx_lm.generate --model hermitdave/Spark-X2.6-1.7B-MLX-slq-4.5bpw --prompt "Hello"

Use `mlx_lm.chat` with the same --model for an interactive chat.

MLX · OpenAI-compatible serverdocs
mlx_lm.server --model hermitdave/Spark-X2.6-1.7B-MLX-slq-4.5bpw --port 8080

API on http://localhost:8080/v1.

Ollama · Q8_0 straight from Hugging Facedocs
ollama run hf.co/XHToken/Spark-X2.6-1.7B-GGUF:Q8_0
llama.cpp · OpenAI-compatible server with a web UIdocs
llama-server -hf XHToken/Spark-X2.6-1.7B-GGUF:Q8_0

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

LM Studio · Download with the lms CLIdocs
lms get XHToken/Spark-X2.6-1.7B-GGUF@Q8_0

Then load it from the LM Studio app or with `lms load`.

Transformers · Chat with the pipeline APIdocs
from transformers import pipeline

pipe = pipeline("text-generation", model="XHToken/Spark-X2.6-1.7B", dtype="auto", device_map="auto")
messages = [{"role": "user", "content": "Explain self-hosting in one paragraph."}]
out = pipe(messages, max_new_tokens=512)
print(out[0]["generated_text"][-1]["content"])

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

All that fit →
ModelParamsBest quantMemorySpeedLicence
LiquidAI logoLFM3-1.7BLiquidAI
1.7BQ8_0≈2.7 GBFits in memory~36 tok/sConditions
1.5BQ8_0≈2.6 GBFits in memory~40 tok/sCommercial OK
2BQ8_0≈3.2 GBFits in memory~31 tok/sCommercial OK
1.2BQ8_0≈2.0 GBFits in memory~50 tok/sConditions
1BQ8_0≈1.8 GBFits in memory~62 tok/sConditions
2.5BQ8_0≈3.5 GBFits in memory~25 tok/sCommercial OK

Other GPUs that run Spark-X2.6-1.7B

Frequently asked questions

Can the Apple M4 (16 GB) run Spark-X2.6-1.7B?
Yes. The Apple M4 (16 GB) runs Spark-X2.6-1.7B fully in unified memory at FP16 / BF16 (about 4.3 GB of 11 GB), at roughly 19 tokens/s.
How fast does Spark-X2.6-1.7B run on the Apple M4 (16 GB)?
An estimate from memory bandwidth gives about 19 tokens/s at FP16 / BF16.
Which quantisation of Spark-X2.6-1.7B should I use on the Apple M4 (16 GB)?
FP16 / BF16: the largest one that fits. Expected quality: 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.