Run Qwen3.5 4B on a Mac
Our docs' default serve example and the most-pulled alias in the catalog — partly because every guide starts here, and it holds up: strong tool calling for its size, 61 tok/s on an M2 Pro, and 3.4 GB resident leaves room for everything else.
4 B dense · 4-bit first model to try · tool calling on small hardware
One command
$ rapid-mlx serve qwen3.5-4b-4bit
Weights (2.9 GB) download on first run; you get an OpenAI-compatible
endpoint at http://localhost:8000/v1. No rapid-mlx yet? It's one line —
curl -fsSL https://rapidmlx.com/install.sh | bash — or take the
desktop app.
Measured on real hardware
| Machine | Decode | First token | Peak memory | Cold boot | Weights |
|---|---|---|---|---|---|
| Mac mini M2 Pro · 32 GB | 60.7 tok/s | 0.56 s | 3.4 GB | 6.0 s | 2.9 GB |
| Mac Studio M3 Ultra · 256 GB † | 158 tok/s | — | 2.4 GB | — | — |
Measured on rapid-mlx 0.12.10, 2026-08-11. Decode: median of 3 runs, temperature 0, 256-token saturating generation, engine-reported token counts, unique salt per request (no prefix-cache hits). TTFT: median of 3, short prompt. Boot: process spawn to first completed token. Peak RSS: 0.5 s sampling across the run. M3 Ultra row: from our 17-model benchmark post (same alias and quant).
Will it fit your Mac?
Peak resident memory measured 3.4 GB during a 256-token generation. The KV cache grows with context length, so treat that as a floor, not a ceiling. For the conservative install-default placement see the hardware tiers table; to compare against every model your RAM can hold, use the live picker.
Variants & alternatives
qwen3.5-9b-4bit— Qwen3.5 9B — the same line, noticeably smarter
FAQ
How much memory does Qwen3.5 4B need on a Mac?
Measured peak resident memory was 3.4 GB on rapid-mlx 0.12.10 during a 256-token generation (M2 Pro, 32 GB). The weights are 2.9 GB on disk. Longer contexts grow the KV cache beyond this, so leave headroom.
How fast is Qwen3.5 4B on Apple Silicon?
We measured 60.7 tokens/sec sustained decode and 0.56 s time-to-first-token on a Mac mini M2 Pro (32 GB), median of 3 runs at temperature 0, and 158 tokens/sec on an M3 Ultra in our published 17-model benchmark.
How do I run Qwen3.5 4B locally?
Install rapid-mlx (curl -fsSL https://rapidmlx.com/install.sh | bash, or brew install rapid-mlx), then: rapid-mlx serve qwen3.5-4b-4bit — the weights download on first run and you get an OpenAI-compatible endpoint at localhost:8000/v1 that works with Cursor, Claude Code, Aider, and any OpenAI client.