How fast does Muse Spark Open run on the Apple M3 Ultra?

Home › Models › Muse Spark Open › Apple M3 Ultra

Muse Spark Open on the Apple M3 Ultra

Muse Spark Open runs at about 18.0 tokens/s on the Apple M3 Ultra (measured, Q2_K, Metal, single-stream llama-server decode).

Best measured generation18.0tok/s

Q2_K · Metal

The source gives no prompt-processing figure.

For comparison, our size-scaling estimate for this combination is 14.1 tok/s.

Every run

Quant	Backend	Generate tok/s	Prompt tok/s	Test	Build / by / notes	Source
Q2_K	Metal	18.0	–	decode, llama.cpp server	512 GB M3 Ultra, ggml-org imatrix Q2_K GGUF on llama.cpp; prompt length not stated.	Hugging Face blog: Welcome Muse Spark Open

tg128 = generating 128 tokens; pp512 = reading a 512-token prompt (llama-bench). LocalScore averages several prompt and output lengths, so its numbers run lower than tg128. "decode, llama-server" = single-stream generation speed measured against a running llama-server (no speculative decoding).

Check prices

Amazon

As an Amazon Associate I earn from qualifying purchases.

No card yet?

Try it on a rented GPU (RunPod)

Affiliate link: if you sign up, RigBench may earn a referral credit. It never affects rankings or numbers.

All results for the Apple M3 Ultra · Muse Spark Open on other hardware

Embed this result

HTML

<a href="https://rigbench.pages.dev/model/muse-spark-open-816b-a42b/on/apple-m3-ultra/"><img src="https://rigbench.pages.dev/badge/muse-spark-open-816b-a42b--apple-m3-ultra.svg" alt="Muse Spark Open · Apple M3 Ultra: 18.0 tok/s (Q2_K, measured)"></a>

Markdown (GitHub README, forums)

[![Muse Spark Open · Apple M3 Ultra: 18.0 tok/s (Q2_K, measured)](https://rigbench.pages.dev/badge/muse-spark-open-816b-a42b--apple-m3-ultra.svg)](https://rigbench.pages.dev/model/muse-spark-open-816b-a42b/on/apple-m3-ultra/)