What is the best local model for coding on the Apple M5?

On the Apple M5, openai/gpt-oss-20b ranks first for coding in our benchmarks (73.0/100 at ~45.3 tok/s).

What is the best local model for agentic workflows on the Apple M5?

On the Apple M5, prism-ml/bonsai-27b ranks first for agentic workflows in our benchmarks (90.3/100 at ~14.3 tok/s). liquid/lfm2-24b-a2b is faster (~70.8 tok/s) and still scores well (89.4/100) — a good pick if you'd rather trade a little quality for speed.

Best Local LLM Performance for Apple M5

Hardware Overview

GPU-specific specifications for local LLM planning.

Apple M5

VRAM: 24GB

Recommended Models

Models we recommend for Apple M5.

Fastest

liquid/lfm2-24b-a2b

24B

70.8 tok/s

Highest token generation speed at 70.8 tok/s.

Best quality

prism-ml/bonsai-27b

27B

14.3 tok/sQuality 81.1

Highest completed quality score at 81.1.

Frequently asked questions

Which models actually run best on the Apple M5, by task, from community benchmark data.

What is the best local model for coding on the Apple M5?: On the Apple M5, openai/gpt-oss-20b ranks first for coding in our benchmarks (73.0/100 at ~45.3 tok/s).
What is the best local model for agentic workflows on the Apple M5?: On the Apple M5, prism-ml/bonsai-27b ranks first for agentic workflows in our benchmarks (90.3/100 at ~14.3 tok/s). liquid/lfm2-24b-a2b is faster (~70.8 tok/s) and still scores well (89.4/100) — a good pick if you'd rather trade a little quality for speed.

Benchmark Results

Completed quality-ranked benchmark results for Apple M5.

Open run page →