← Canonical rankings

M4 Pro — benchmark record

M4 Pro local LLM benchmarks across 24 GB, 32 GB, 48 GB, 64 GB RAM tiers on Apple Silicon Mac. 24 published rows across 21 models with explicit evidence state and RAM-tier comparison. Peak published speed is 118.0 tok/s.

24Benchmark rows
21Models tested
4RAM configurations
118.0Fastest avg tok/s

Each configuration differs only in unified memory. More RAM = larger models fit. Throughput is similar across RAM tiers at the same model size.

All benchmark rows — M4 Pro

Sorted by avg tok/s descending. Click source badge to see original measurement.

Chip (RAM)ModelQuantRAM req.Avg tok/sPrompt tok/sRuntimeSource
M4 Pro (24 GB)Qwen 3 4BQ4_K - Medium—118.0 tok/s—MLXref
M4 Pro (24 GB)Phi-4 Mini Instruct 3.8BQ4_K - Medium—108.0 tok/s—Ollamaref
M4 Pro (24 GB)Mistral 7B v0.3Q4_K - Medium—98.0 tok/s—MLXref
M4 Pro (24 GB)Gemma 4 E2BQ4_K - Medium—95.0 tok/s—Ollamaref
M4 Pro (24 GB)Gemma 3 4BQ8_0—95.0 tok/s—MLXref
M4 Pro (24 GB)Qwen3.5-9BQ4_K - Medium—92.0 tok/s—MLXref
M4 Pro (24 GB)Qwen 3 8BQ4_K - Medium—82.0 tok/s—Ollamaref
M4 Pro (24 GB)Gemma 4 E4BQ4_K - Medium—78.0 tok/s—MLXref
M4 Pro (24 GB)Qwen 3 8BQ8_0—68.0 tok/s—MLXref
M4 Pro (24 GB)Mistral 7B v0.3Q8_0—65.0 tok/s—Ollamaref
M4 Pro (48 GB)Qwen 3 30B-A3B8bit—55.0 tok/s—MLXref
M4 Pro (24 GB)Gemma 3 12BQ4_K - Medium—52.0 tok/s—MLXref
M4 Pro (24 GB)Ministral 3 14BQ4_K - Medium—40.0 tok/s—Ollamaref
M4 Pro (24 GB)Qwen 3 14BQ4_K - Medium—38.0 tok/s—LM Studioref
M4 Pro (24 GB)Qwen 3 30B-A3BQ4_K - Medium—35.0 tok/s—MLXref
M4 Pro (24 GB)Qwen3.6-35B-A3BQ4_K - Medium—32.0 tok/s—Ollamaref
M4 Pro (24 GB)Gemma 4 26B-A4BQ4_K - Medium—28.0 tok/s—Ollamaref
M4 Pro (24 GB)Gemma 3 27BQ4_K - Medium—25.0 tok/s—LM Studioref
M4 Pro (24 GB)Nemotron Cascade 2 30B-A3BQ4_K - Medium—22.0 tok/s—Ollamaref
M4 Pro (32 GB)Qwen 3 32BQ4_K - Medium—15.0 tok/s—Ollamaref
M4 Pro (24 GB)Gemma 4 31BQ4_K - Medium—14.0 tok/s—Ollamaref
M4 Pro (48 GB)Devstral Small 1.16bit18.5 GB12.9 tok/s—LM Studioref
M4 Pro (48 GB)Qwen3.5-27B8bit—8.5 tok/s—MLXref
M4 Pro (64 GB)Llama 3.3 70BQ4_K - Medium—5.0 tok/s—Ollamaref

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 →