M3 Ultra (256 GB) — benchmark record

M3 Ultra (256 GB) local LLM benchmarks on Apple Silicon Mac with 13 published rows across 10 models. Peak published speed is 370.0 tok/s on Qwen 3 0.6B. Largest published fit uses 129.8 GB. Published runtimes include MLX.

13Benchmark rows

10Models tested

370.0Fastest avg tok/s (Qwen 3 0.6B)

0Lab benchmarks

Quick take

Best published speed here is 370.0 tok/s on Qwen 3 0.6B at 4bit. Largest published memory footprint is Qwen3.5-122B-A10B at 8bit, using 129.8 GB. This page is evidence, not the full buying answer.

Based on 13 external benchmarks; no lab runs yet.

Published runtimes: 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 Qwen 3 0.6B, Qwen3.5-9B, Qwen3.5-35B-A3B, Qwen3-Coder-Next plus 6 more published models. Published runtime coverage on this chip includes MLX. Fastest published row is 370.0 tok/s on Qwen 3 0.6B at 4bit. Largest published footprint on this page is 129.8 GB for Qwen3.5-122B-A10B.

Qwen 3 0.6B Qwen3.5-9B Qwen3.5-35B-A3B Qwen3-Coder-Next GLM-4.7-Flash Qwen3.5-122B-A10B

Other published M3 Ultra variants: 512 GB

Macs shipping with the M3 Ultra (256 GB)

Mac Studio M3 Ultra 96GB Mac Studio M3 Ultra 256GB

This chip is part of a family. Compare all M3 Ultra RAM variants →

Raw benchmark rows for M3 Ultra (256 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
Qwen 3 0.6B	4bit	—	—	370.0 tok/s	—	MLX	ref
Qwen3.5-9B	4bit	5.1 GB	—	106.0 tok/s	—	MLX	ref
Qwen3.5-35B-A3B	4bit	19.6 GB	—	95.0 tok/s	—	MLX	ref
Qwen3.5-35B-A3B	8bit	36.9 GB	—	80.0 tok/s	—	MLX	ref
Qwen3-Coder-Next	4bit	44.9 GB	—	74.0 tok/s	—	MLX	ref
Qwen3-Coder-Next	6bit	64.8 GB	—	66.0 tok/s	—	MLX	ref
GLM-4.7-Flash	8bit	—	—	58.0 tok/s	—	MLX	ref
Qwen3.5-122B-A10B	MXFP4	65.0 GB	—	57.0 tok/s	—	MLX	ref
GLM-4.5-Air	4bit	60.3 GB	—	54.0 tok/s	—	MLX	ref
Devstral Small 2 24B	4bit	13.4 GB	—	47.0 tok/s	—	MLX	ref
Qwen3.5-122B-A10B	8bit	129.8 GB	—	43.0 tok/s	—	MLX	ref
Qwen3.5-27B	4bit	15.3 GB	—	38.0 tok/s	—	MLX	ref
Qwen 3 235B-A22B	Q4	—	—	27.0 tok/s	—	MLX	ref

Models tested on this chip

Qwen 3 0.6B Qwen3.5-9B Qwen3.5-35B-A3B Qwen3-Coder-Next GLM-4.7-Flash Qwen3.5-122B-A10B GLM-4.5-Air Devstral Small 2 24B Qwen3.5-27B Qwen 3 235B-A22B

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 from in-house lab measurements plus community-published benchmarks. See all chip families →