Silicon Score Apple Silicon local LLM rankings
← Canonical rankings

Silicon Score › Chips › M3 (16 GB)

M3 (16 GB) — benchmark record

M3 (16 GB) local LLM benchmarks on Apple Silicon Mac with 9 published rows across 9 models. Peak published speed is 95.0 tok/s on Phi-4 Mini Instruct 3.8B. Published runtimes include LM Studio, MLX, Ollama.

9Benchmark rows
9Models tested
95.0Fastest avg tok/s (Phi-4 Mini Instruct 3.8B)
0Lab benchmarks

Quick take

Best published speed here is 95.0 tok/s on Phi-4 Mini Instruct 3.8B at Q4_K - Medium. This page is evidence, not the full buying answer.

Based on 9 external benchmarks; no lab runs yet.

Published runtimes: LM Studio, MLX, Ollama.

Need a recommendation? Use BuyChecking a target model? Use FitWant a setup answer? Use RunBrowse Macs by exact hardwareNeed the full audit trail? Use Bench

Current published coverage

Published model coverage includes Phi-4 Mini Instruct 3.8B, Gemma 4 E2B, Gemma 4 E4B, Mistral 7B v0.3 plus 5 more published models. Published runtime coverage on this chip includes LM Studio, MLX, Ollama. Fastest published row is 95.0 tok/s on Phi-4 Mini Instruct 3.8B at Q4_K - Medium.

Phi-4 Mini Instruct 3.8BGemma 4 E2BGemma 4 E4BMistral 7B v0.3Qwen3.5-9BMinistral 3 8B

Raw benchmark rows for M3 (16 GB)

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

ModelQuantRAM req.ContextAvg tok/sPrompt tok/sRuntimeSource
Phi-4 Mini Instruct 3.8BQ4_K - Medium——95.0 tok/s—MLXref
Gemma 4 E2BQ4_K - Medium——82.0 tok/s—Ollamaref
Gemma 4 E4BQ4_K - Medium——62.0 tok/s—Ollamaref
Mistral 7B v0.3Q4_K - Medium——62.0 tok/s—Ollamaref
Qwen3.5-9BQ4_K - Medium——58.0 tok/s—Ollamaref
Ministral 3 8BQ4_K - Medium——55.0 tok/s—LM Studioref
Gemma 3 12BQ4_K - Medium——38.0 tok/s—Ollamaref
Qwen 3 14BQ4_K - Medium——30.0 tok/s—Ollamaref
Ministral 3 14BQ4_K - Medium——30.0 tok/s—LM Studioref

Models tested on this chip

Phi-4 Mini Instruct 3.8BGemma 4 E2BGemma 4 E4BMistral 7B v0.3Qwen3.5-9BMinistral 3 8BGemma 3 12BQwen 3 14BMinistral 3 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 →