Fastest
gemma-4-26B-A4B-it-MLX-8bit
26B
68.8 tok/s
Highest token generation speed at 68.8 tok/s.
Hardware Benchmarks
Best Local LLM Performance for Apple M2 Ultra
Compare Apple M2 Ultra benchmark results and see which models actually earn the best completed quality scores.
Hardware Overview
GPU-specific specifications for local LLM planning.
- VRAM
- 128GB
Recommended Models
Models we recommend for Apple M2 Ultra.
Best quality
Qwen3.8-27B-oQ8e-fp16-mtp
27B
34.6 tok/sQuality 86.6
Highest completed quality score at 86.6.
Frequently asked questions
Which models actually run best on the Apple M2 Ultra, by task, from community benchmark data.
- What is the best local model for coding on the Apple M2 Ultra?
- On the Apple M2 Ultra, Qwen3.8-27B-oQ8e-fp16-mtp ranks first for coding in our benchmarks (72.7/100 at ~34.1 tok/s).
- What is the best local model for agentic workflows on the Apple M2 Ultra?
- On the Apple M2 Ultra, Qwen3.8-Flash-Next-oQ4e-mtp ranks first for agentic workflows in our benchmarks (87.0/100 at ~25.2 tok/s). Qwen3.6-27B-oQ8-mtp is faster (~32.8 tok/s) and still scores well (80.0/100) — a good pick if you'd rather trade a little quality for speed.
Benchmark Results
Completed quality-ranked benchmark results for Apple M2 Ultra.