How much RAM do you need on a Mac for development?

The short answer

Memory	Comfortable for
8 GB	Browsing, writing, a single small web project. Swaps often with a modern dev setup.
16 GB	Most web and mobile development: an editor, a browser, a couple of dev servers, light Docker.
24 GB	The same with room to spare: Docker with a few services, Xcode plus simulators, small local models.
32 GB	Heavy Docker stacks, several projects open, Android emulator plus iOS simulator, 7B to 14B local models alongside your tools.
64 GB+	Local LLMs in the 30B to 70B range, big data work, multiple VMs.

These assume you also keep a browser, Slack and a few other apps open, because everyone does.

What actually uses memory in a dev setup

Rough footprints you’ll see in Activity Monitor’s Memory column, added up across helpers:

• Browser with 20 to 40 tabs: 2 to 6 GB.
• VS Code or Cursor with a TypeScript project: 1 to 4 GB, more with many extensions or a big monorepo.
• Slack, Discord, Notion, Figma: 300 MB to 1.5 GB each. They’re Electron or browser-based.
• Docker Desktop: its VM can use up to the memory limit in Settings › Resources. A Postgres, Redis and app container stack often takes 2 to 6 GB.
• Xcode and a simulator: 3 to 8 GB together while building.
• Android Studio and an emulator: 4 to 8 GB.

Add your own numbers. If they come to 14 GB before you’ve started a build, 16 GB will swap.

Local LLMs change the math

A model’s weights have to fit in memory. For the common 4-bit quantized versions, a rough rule is a bit over half a gigabyte per billion parameters, plus room for context:

Model size (4-bit)	Approximate memory
7B to 8B	5 to 6 GB
13B to 14B	9 to 10 GB
30B to 34B	18 to 22 GB
70B	40 GB or more

Longer context windows and higher-precision versions need more. macOS also caps how much unified memory the GPU can use by default (a large share of it, but not all), so a model that technically fits may still be slow or fall back to the CPU. If you want to run 30B-class models while also coding, 48 to 64 GB is where that becomes comfortable.

Let your current Mac decide

Specs tables are guesses. Your own workload is data. For a week, check memory pressure during your normal day:

1. Open Activity Monitor › Memory and keep an eye on the Memory Pressure graph.
2. Note Swap Used at the end of each day:

sysctl vm.swapusage

3. During your heaviest moment (a build, a Docker stack, a call with screen share), check whether apps hesitate when you switch.

Then:

What you see	What it means
Green almost all day, swap under 2 GB	You have enough. Buy the same again, or one step up for headroom.
Yellow during builds, green otherwise	You’re at the edge. One step up (16 → 24, 24 → 32) will feel noticeably better.
Yellow most of the day, swap growing past 5 to 10 GB	You need more. Aim for roughly double what you have.
Red regularly	Your workload is well past your RAM. Double at least.

The colors are explained in memory pressure on Mac, and what growing swap means is in high swap usage on Mac.

Memory or storage?

If budget forces a choice, developers usually feel more RAM sooner than more storage. Storage can be extended with a fast external drive; RAM can’t be. The exception is a 256 GB drive with Docker and Xcode, which fills fast. Also, a nearly full disk leaves less room for swap, which makes a low-memory Mac worse.

Seeing it live

Stat Space keeps memory pressure in your menu bar and saves per-minute averages of memory, pressure, swap and paging for 30 days. Instead of checking Activity Monitor every few hours, open the history chart at the end of the week and see how often you were in yellow and how high swap went. Pro’s Trends tab reads the same history back in plain words, such as time spent with pressure elevated and your swap peak. History is free; Trends is Pro.