Sunk Cost sunkcost.ai Data checked 2026-09-03

Mac Studio M5 Max, 36GB vs 48GB for local AI

Both hold 27 of the 39 open models here. The Mac Studio M5 Max, 36GB costs $600 less. On Qwen3.8 27B, the strongest model both hold, the Mac Studio M5 Max, 48GB is about 1.3× faster: 25 tok/s against 19, both estimated from memory bandwidth. The Mac Studio M5 Max, 36GB pays for itself sooner, in 32 years against 39 years at 500k tokens a day.

Mac Studio M5 Max, 36GBMac Studio M5 Max, 48GB
Price$2,499$3,099
Memory36 GB48 GB
Usable by the GPU27 GB36 GB
Memory bandwidth460 GB/s614 GB/s
Power under load145 Wstand-in145 Wstand-in
Models that fit2727
Best model it runsQwen3.8 27BQwen3.8 27B
Speed on that model19 tok/s estimated25 tok/s estimated
Pay-back on that modelPays back in 32 yearsPays back in 39 years

Neither power figure above is measured on the machine beside it. Both are stand-ins borrowed from the nearest hardware the data does have, so what the row shows is one borrowed figure printed twice. Each machine's page names the figure it borrows and why. Every pay-back figure on this page prices its electricity from these numbers.

Run the numbers on the Mac Studio M5 Max, 36GB · or the Mac Studio M5 Max, 48GB

The step up is a different chip, not just more memory

Mac Studio M5 Max is one name for two machines. The data lists the 36GB as 18-core CPU / 32-core GPU and the 48GB as 18-core CPU / 40-core GPU. The step adds 8 GPU cores to the graphics part, 40 against 32. That is why this pair is not one of the site's memory-size comparisons: those hold the silicon equal so the memory is the whole of the difference, and Apple cut both to reach the $2,499.

The dearer chip has the wider path to memory as well, 614 GB/s against 460. Writing a token means reading the whole model out of memory, so that is the figure the speeds in the table follow, and it is the part of this step you can see in them. Both hold the same 27 models at 32k of context. Price the Mac Studio M5 Max, 36GB on its own before paying for the step: here the step buys room and speed together, and the years move with both.

How much use it takes to pay back

Everything above is at 500k tokens a day. Pay-back moves with how much you actually run, so here are both machines on Qwen3.8 27B, the strongest model both hold, at the five levels of use the calculator names. The Mac Studio M5 Max, 36GB pays back sooner at every level of use, so this is not a choice that turns on how hard you work it.

A day's useMac Studio M5 Max, 36GBMac Studio M5 Max, 48GB
50ka few chats a day316 years387 years
200klight assistant use79 years97 years
1Ma moderate coding-assistant day16 years19 years
4Mheavy coding with an agent3.9 years4.8 years
20Magents running most of the day9.5 months12 months

Run the Mac Studio M5 Max, 36GB at 20M tokens a day · or the Mac Studio M5 Max, 48GB

The same models, not to the same length

Every model on this list that fits one machine fits the other at 32k of context, so memory does not change what they run. What it changes is how far you can take the context on 11 of them. The Mac Studio M5 Max, 48GB has 36 GB usable against 27 GB, and spare memory is what the KV cache grows into as you keep more tokens.

ModelMac Studio M5 Max, 36GBMac Studio M5 Max, 48GB
Qwen3.8 27B16 GB of weights128k256k
Qwen3.6 27B17 GB of weights128k256k
Qwen3.6 35B-A3B22 GB of weights128k256k
Gemma 4 31B it20 GB of weights32k64k
Granite 4.2 30B18 GB of weights32k64k
Nemotron 3.5 Lightning 30B-A3B25 GB of weights128k256k
Qwen3-Coder 30B-A3B19 GB of weights64k128k
Devstral Small 2 24B14 GB of weights64k128k
Ministral 3 14B8.2 GB of weights64k128k
Laguna XS 2.120 GB of weights128k256k
Ornith 1.5 35B-A3B22 GB of weights128k256k

Each figure is the longest context the calculator offers that the machine still holds that model at, and no model is taken past its own context limit. The other 28 models the calculator counts reach the same length on both machines, at every setting from 4k to 256k.

Run Qwen3.8 27B on the Mac Studio M5 Max, 36GB at 128k · or on the Mac Studio M5 Max, 48GB at 256k

The assumptions behind both columns

Both columns use the same usage: 500k tokens a day at 15:1 input to output, 32k of context, $0.17 per kWh, and today's API prices held flat. Speeds marked estimated are worked out from memory bandwidth rather than measured, and pay-back scales with them. Change any of it in the calculator.

More head to head: everything the Mac Studio M5 Max, 36GB runs · everything the Mac Studio M5 Max, 48GB runs · every other match-up · the quickest pay-back at each level of use · every model against the frontier