Mac LLM Calculator tells you which Apple Silicon runs which model, and how fast
shauntrennery · x · 2026-10-17
Macyou launched a web calculator that tells you whether a given Mac can run a given LLM locally and at what speed, based on two numbers: whether weights fit in unified memory (parameter count + quantization arithmetic) and generation speed (memory bandwidth ÷ weight size).
- Covers M1 through M5/Ultra chips with their bandwidth figures (e.g. M5 Ultra at 1200 GB/s), 24–64GB memory tiers, quantization options from Q4KM to FP16, and a wide model list.
- Example: Qwen3.8 27B at Q4KM on a 64GB M4 Pro runs at roughly 14 tokens/sec, with 16.9GB weights plus 1GB context, and 23.4GB of 64GB taken once macOS is accounted for.
- Also offers monthly Mac mini rental pricing for machines that can run your chosen model.
More from Infra
- Ben Lorica: Local AI Is a Routing Problem, Not Just a Privacy Feature — bigdata · 2026-10-17
- Bittensor expands subnet capacity to 2,500 miners, opening autonomous AI agent participation — bittingthembits · 2026-10-17
- AI optics stocks fell 25-31% in a week, but Taiwan server revenue still doubling — tengyanAI · 2026-10-17
- Hugging Face to livestream speculative decoding tutorial for speeding up local LLMs — huggingface · 2026-10-17
- David Sacks pushes back: CoreWeave bond-yield spike reflects AI volatility, not demand collapse — DavidSacks · 2026-10-17
- Custom VBIOS loader lets you edit RTX 50/40/30 power limits and voltages without flashing — panchovix · 2026-10-17