Local LLM on Apple M4 Max

Is the Apple M4 Max the right card for this?

On Apple Silicon the model shares system memory with everything else, so 36 GB is a working budget rather than a hard wall. 410 GB/s at 40 W is 10.3 GB/s per watt, which is why a laptop can hold a conversation with a mid-sized model on battery. Discrete cards in the same price bracket are faster per token but cannot be carried anywhere. At around $2400 that is about $67 per gigabyte. Catalogue throughput class: Q4 7B: 60-80, Q4 32B: 22-30, Q4 70B: 7-10.

A 2024 part sits in the middle of this guide's range: old enough that backend support has stopped moving and the community has written up the quirks, new enough that nothing treats it as legacy. That is the least risky place on the calendar to buy from if you want the setup to work the first time.

What the price gap buys

No other card in this guide carries 36 GB, so there is no like-for-like price comparison to make. The nearest capacities are listed at the foot of this page, and moving to any of them changes which models fit, not just how fast they run.

At 2 years old this is still a current retail part, so the price above is a shop price rather than a listing price and stock is the usual constraint rather than condition. Buying new also means the 40 W figure is the number your power supply has to meet from day one.

Where the Apple M4 Max sits in this guide

Across the 20 cards catalogued here, the Apple M4 Max ranks 15th on memory bandwidth at 410 GB/s, 13th cheapest at about $2400, and 3rd on bandwidth per watt at 10.3 GB/s per watt. Those three positions, not the capacity figure, are what separate it from other 36 GB parts.

Unified LPDDR5X is laptop-class memory shared between CPU and GPU, so there is no copy step between "system RAM" and "video RAM" and no fixed split between them. 410 GB/s is modest next to discrete graphics memory, but the model never has to be moved, and the whole 36 GB is available to it.

On Apple hardware the fast path is MLX, Apple's own array framework, with Ollama as the simpler alternative running through Metal. Our card records for Apple parts cite mlx-lm benchmarks for exactly this reason: the same machine gives materially different numbers depending on which of the two you use.

Which catalogue models fit on Apple M4 Max?

Of the 46 models in our catalogue with per-size memory figures, 39 have at least one size that fits in 36 GB at Q4_K_M with room for context. The largest of each are below, biggest first.

Model	Largest size that fits	Weights at Q4_K_M
LLaVA	34B	22 GB
Command R	35B	22 GB
Qwen 2.5	32B	20 GB
Qwen 2.5-Coder	32B	20 GB
Qwen 3	32B	20 GB
DeepSeek R1	32B	20 GB
DeepSeek Coder	33B	20 GB
Gemma 4	31b	20 GB

A further 31 smaller models also fit; see the model catalogue.

The next sizes up are out of reach without offloading: Llama 3.1 70B at 42 GB, Llama 3.3 70B at 42 GB, DeepSeek R1 70B at 42 GB.

Expected tokens per second on Apple M4 Max

Memory bandwidth, not compute, is the bottleneck for single-stream token generation. The Apple M4 Max's 410 GB/s puts the following ceiling on a batch of one at 2048 tokens of context.

Model	Quantization	Status	Approx tokens/sec
Llama 3.2 1B	Q4_K_M	Fits in VRAM	~74 tok/s
Llama 3.1 8B	Q4_K_M	Fits in VRAM	~25 tok/s
Llama 3.1 8B	Q8_0	Fits in VRAM	~18 tok/s
Mistral Nemo 12B	Q4_K_M	Fits in VRAM	~16 tok/s
Qwen 2.5 14B	Q4_K_M	Fits in VRAM	~14 tok/s
Qwen 3 32B	Q4_K_M	Fits in VRAM	~7 tok/s
DeepSeek R1 distilled 32B	Q4_K_M	Fits in VRAM	~7 tok/s
Llama 3.1 70B	Q4_K_M	Needs CPU offload	~1 tok/s

These come from a bandwidth heuristic (tokens/sec ≈ bandwidth in GB/s × a per-model efficiency factor), not from a test rig in our office. Real numbers move with model architecture, batch size and KV-cache size. Treat them as an order of magnitude, and treat the "reported throughput class" row in the table above as the community-measured figure.

Build recommendations for Apple M4 Max

Best model to download first

Llama 3.1 8B for everyday chat and coding; a 32B at Q4_K_M for top quality within 36 GB. Start with ollama pull llama3.1:8b.

Recommended inference backend

For Apple Silicon, use MLX (best performance) or Ollama (simplest setup). Mullama works via llama.cpp but won't beat MLX on Apple Silicon.

Sources

[1] apple.com/macbook-pro · [2] mlx-lm benchmarks · VRAM figures for catalogue models from src/data/models.json (2026-06-29); card specifications from src/data/gpus.json (2026-06-29).