# Apple M5 Max

53 Hacker News commenters recommended Apple M5 Max and 17 criticized it, November 2025 to November 2026. It was recommended in 13 of 13 months. It meets the Hacker Approved badge rule.

**Verdict:** Praised for strong local LLM inference from large unified memory, fast CPU and GPU speed, and quiet, efficient performance. Critics agree memory bandwidth is the main limit for inference.

## Why people recommend it

- Strong local LLM inference (6 comments): "M5 Max is a major improvement for local inference due to the added matmul accelerators, but the RAM capacity and bandwidth bottleneck is huge." ([source](https://news.ycombinator.com/item?id=48067926)) "You can already do this. Get an M5 Max with 129GB model and run a Qwen model." ([source](https://news.ycombinator.com/item?id=47469268)) "runs llms surprisingly well without quantization, at least the moe models" ([source](https://news.ycombinator.com/item?id=48201501))
- Better prompt processing than before (3 comments): "it will fix the biggest issue with the current lineup, prompt processing, in addition to a bandwidth bump" ([source](https://news.ycombinator.com/item?id=46362787)) "at 16k pre-fill, the M5 Max is 3.6 times faster than the M4 Max" ([source](https://news.ycombinator.com/item?id=48162564)) "m5 max really fixed pp with the better matmul support" ([source](https://news.ycombinator.com/item?id=49530348))
- Fast CPU and GPU performance (4 comments): "in Blender, the M4 Max is on par with the top of Nvidia's (mobile) 5xxx line, except with much more VRAM" ([source](https://news.ycombinator.com/item?id=46802761)) "one of the best laptop compute GPUs for blender rendering" ([source](https://news.ycombinator.com/item?id=48902315)) "Apple CPU holds the fastest single core CPU benchmarks" ([source](https://news.ycombinator.com/item?id=46934222))
- Quiet and power efficient (3 comments): "the acoustic engineers at Apple have done a very good job of making it be pretty much not noticeable" ([source](https://news.ycombinator.com/item?id=49124671)) "can almost match it in the CPU-bound workload I care about while sipping some 45W" ([source](https://news.ycombinator.com/item?id=47865381)) "M5 max is 3x stronger and 50% more power efficient" ([source](https://news.ycombinator.com/item?id=48353509))

## What people warn about

- Memory bandwidth too low for inference (4 comments): "M5 Max is a major improvement for local inference due to the added matmul accelerators, but the RAM capacity and bandwidth bottleneck is huge." ([source](https://news.ycombinator.com/item?id=48067926)) "that chip just doesn't really have the memory bandwidth to run this (or nearly any) model at acceptable speeds" ([source](https://news.ycombinator.com/item?id=49300012)) "my only concern has been the very meager memory bandwidth of these chips" ([source](https://news.ycombinator.com/item?id=49273670))
- Not competitive for compute workloads (4 comments): "the M5 Max still doesn't have the compute to make it practical, at least from an interactive usage standpoint" ([source](https://news.ycombinator.com/item?id=48429811)) "the fastest Apple Silicon chips like the M5 Max and the M3 Ultra still put up worse GPU compute performance" ([source](https://news.ycombinator.com/item?id=49305582)) "They're honestly not competitive for inference, it's why datacenters largely ignore Apple Silicon." ([source](https://news.ycombinator.com/item?id=48390576))
- Slow prefill and time to first token (3 comments): "It's feasible to run 256k token workflows that would take the better half of an hour for TTFT on the M5." ([source](https://news.ycombinator.com/item?id=48430354)) "My biggest problem with running local LLMs on my M4 Max/128GB RAM is the prefill latency." ([source](https://news.ycombinator.com/item?id=49530249)) "the M5 Max has a 50% longer TTFT on Qwen's 27b dense model at only 16k of context" ([source](https://news.ycombinator.com/item?id=48864460))
- High price for many buyers (2 comments): "which isn't very "enthusiast-compatible", it's about 3x the minimum salary in my country!" ([source](https://news.ycombinator.com/item?id=47542509)) "I do not see Apple as a competitive solution, as their price is much, much higher" ([source](https://news.ycombinator.com/item?id=48808564))

## Alternatives hackers recommend

- [Apple M1 laptop](https://hackerapproved.com/p/apple-m1-chip.md): 58 recommend
- [Snapdragon X Elite](https://hackerapproved.com/p/snapdragon-x-elite.md): 34 recommend
- [Intel Panther Lake](https://hackerapproved.com/p/intel-panther-lake.md): 32 recommend
- [Intel Lunar Lake](https://hackerapproved.com/p/intel-lunar-lake.md): 17 recommend

## Where to get it

- [See price on Amazon](https://www.amazon.com/s?k=Apple+MacBook+Pro+M5+Max+128GB&tag=hackerapprove-20) (affiliate link)
- [Official site](https://www.apple.com/macbook-pro/)

## Cite as

Hacker Approved, "Apple M5 Max", https://hackerapproved.com/p/apple-m5-max/ (counts from 1,795,640 Hacker News comments, November 2025 to November 2026).