M4 Ultra (192 GB) — benchmark record

M4 Ultra (192 GB) local LLM benchmarks on Apple Silicon Mac with 5 published rows across 5 models. Peak published speed is 45.0 tok/s on Mistral Small 4 119B. Published runtimes include MLX. Evidence state and next-step ranking links included.

5Benchmark rows

5Models tested

45.0Fastest avg tok/s (Mistral Small 4 119B)

0Silicon Score Lab rows

What this page answers best

Best published speed here is 45.0 tok/s on Mistral Small 4 119B at Q4_K - Medium. This page is evidence, not the full buying answer.

Evidence state: 5 linked reference rows and no Silicon Score Lab rows yet.

Published runtimes here: MLX.

Need a recommendation? Use Buy Checking a target model? Use Fit Want a setup answer? Use Run Browse Macs by exact hardware Need the full audit trail? Use Bench

Current published coverage

Published model coverage includes Mistral Small 4 119B, Llama 4 Scout 17B-16E, Qwen 3 235B-A22B, Llama 3.3 70B plus 1 more published model. Published runtime coverage on this chip includes MLX. Fastest published row is 45.0 tok/s on Mistral Small 4 119B at Q4_K - Medium.

Mistral Small 4 119B Llama 4 Scout 17B-16E Qwen 3 235B-A22B Llama 3.3 70B gpt-oss 120B

Raw benchmark rows for M4 Ultra (192 GB)

Rows sorted by avg tok/s descending. Click source badge to see original measurement page.

Model	Quant	RAM req.	Context	Avg tok/s	Prompt tok/s	Runtime	Source
Mistral Small 4 119B	Q4_K - Medium	—	—	45.0 tok/s	—	MLX	ref
Llama 4 Scout 17B-16E	Q4_K - Medium	—	—	30.0 tok/s	—	MLX	ref
Qwen 3 235B-A22B	Q4_K - Medium	—	—	22.0 tok/s	—	MLX	ref
Llama 3.3 70B	Q4_K - Medium	—	—	18.0 tok/s	—	MLX	ref
gpt-oss 120B	Q4_K - Medium	—	—	10.0 tok/s	—	MLX	ref

Models tested on this chip

Mistral Small 4 119B Llama 4 Scout 17B-16E Qwen 3 235B-A22B Llama 3.3 70B gpt-oss 120B

Next step

Use Rankings when you need the best overall answer for a Mac, and keep this page open when you need the evidence behind that recommendation.

If you are comparing local Apple Silicon against rented cloud hardware, use AI Data Center Index for current GPU rental context.

Data

benchmarks.json — full dataset · chips.json — chip summaries · benchmarks.csv — CSV export

Data sourced from Silicon Score Lab measurements and community reference runs. See all chip families →