M4 (10-core GPU, 16 GB) — benchmark record

M4 (10-core GPU, 16 GB) local LLM benchmarks on Apple Silicon Mac with 4 published rows across 4 models. Peak published speed is 76.2 tok/s on Llama 3.2 1B. Largest published fit uses 3.6 GB. Published runtimes include llama.cpp, llamafile.

4Benchmark rows

4Models tested

76.2Fastest avg tok/s (llama-3-2-1b-instruct)

0Lab benchmarks

Quick take

Best published speed here is 76.2 tok/s on Llama 3.2 1B at Q4_K - Medium. Largest published memory footprint is Llama 2 7B at Q4_0, using 3.6 GB. Longest published context on this page is 512. This page is evidence, not the full buying answer.

Based on 4 external benchmarks; no lab runs yet.

Published runtimes: llama.cpp, llamafile.

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 Llama 3.2 1B, Llama 2 7B, Llama 3.1 8B, Qwen 2.5 14B. Published runtime coverage on this chip includes llama.cpp, llamafile. Fastest published row is 76.2 tok/s on Llama 3.2 1B at Q4_K - Medium. Largest published footprint on this page is 3.6 GB for Llama 2 7B. Longest published context is 512.

Llama 3.2 1B Llama 2 7B Llama 3.1 8B Qwen 2.5 14B

Other published M4 (10-core GPU) variants: 24 GB · 32 GB

This chip is part of a family. Compare all M4 (10-core GPU) RAM variants →

Raw benchmark rows for M4 (10-core GPU, 16 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
Llama 3.2 1B	Q4_K - Medium	—	—	76.2 tok/s	1091.1 tok/s	llamafile	ref
Llama 2 7B	Q4_0	3.6 GB	512	24.1 tok/s	221.3 tok/s	llama.cpp	ref
Llama 3.1 8B	Q4_K - Medium	—	—	16.0 tok/s	166.8 tok/s	llamafile	ref
Qwen 2.5 14B	Q4_K - Medium	—	—	8.7 tok/s	83.1 tok/s	llamafile	ref

Models tested on this chip

Llama 3.2 1B Llama 2 7B Llama 3.1 8B Qwen 2.5 14B

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 →