M5 32GB

APPLE · 32GB unified memory · 4 reports

On the M5 32GB, Qwen3 30B · 3B active at 4-bit, the model most run on it, generates a median of 57.0 tokens per second from one plain run (community reports, as of 29 Oct 2026).

As of 29 Oct 2026, the models most run on the M5 32GB, with generation and prompt-processing medians from plain runs (one device, one request, no speculative decoding):

Qwen3 30B · 3B active
57.0gen t/s at 4-bit · one run
PP 1063.0 t/s · 1 run

Engines people use on it: MLX 2 (reports) · oMLX 1 (reports)

See reported performance Estimate what fits

How to compare these reports →Filter reports by setup →All M5 Macs compared →

Run models on your M5 32GB? Add your numbers.

Your coding agent can run a speed test on your machine and send it here with a free account. Or paste a result you've already posted.

Benchmark with your agent Paste a result

Reported performance on the M5 32GB

These reports mix GPU counts and memory setups, so a model listed here may have needed extra cards or CPU offloading. The calculator estimates which models fit in 32 GB of unified memory.

Date ↓Date ↑Token speed ↓Token speed ↑Prefill ↓Prefill ↑Context ↓Context ↑Ranked

Qwen3.6 35B (3B active)

M5 32GB · oMLX

generation:: 40.0 tokens/s
kv:: 8-bit

Reported by the source; GPU count, offloading and concurrent requests can change these figures. Prompt-processing speed is shown only when the source reports it; generation speed cannot tell us time to first token. Check the full setup before comparing.

codingagentic

User reports Qwen 3.6 35B-A3B at about 40 tok/s on an M5 MacBook Pro with 32GB unified memory, against about 17 tok/s for a dense 27B at 4-bit. Setup uses oMLX with an SSD-backed cache server, 8-bit KV cache and context set to roughly 32K. The user says the 16GB dense model plus KV cache filled the 32GB machine and caused a shutdown; the MoE has been stable since.

Qwen3 30B (3B active)

M5 32GB · MLX

generation:: 57.0 tokens/s
prompt processing (prefill):: 1063.0 tokens/s
quant:: 4-bit (MLX)

Reported by the source; GPU count, offloading and concurrent requests can change these figures. Prompt-processing speed is input throughput, not time to first token. Check the full setup before comparing.

User reports Qwen3-30B-A3B 4-bit at 57 tok/s cold generation on an M5 MacBook Air with 32GB. Setup is MLX with mlx-lm, 4-bit weights, 256 generated tokens, no speculative decoding. Sustained generation drops to 44 tok/s on battery and 39 tok/s on mains; prompt processing of a 4,096-token prompt is 1,063 tok/s cold, 788 sustained on battery and 725 on mains. The decline finishes at about 150 seconds and the rate then holds flat.

DeepSeek V4 Flash 284B (13B active)

M5 32GB

generation:: 1.0 tokens/s
prompt processing (prefill):: 50.0 tokens/s
quant:: 4bit

User reports DeepSeek V4 Flash 0731 at about 50 t/s prefill and about 1 t/s decode on an M5 Air with 32 GB. Setup uses the streamed experts trick with a model of roughly 300B total parameters.

Gemma 4

M5 32GB · MLX · 130,173 ctx

prompt processing (prefill):: 3029.0 tokens/s

User reports custom w8a8 kernels on an M5 MacBook Air, with prefill improving from 2193 t/s to 3029 t/s for 130k tokens. User also mentions llama.cpp for Macs.

Get a weekly email of new M5 32GB reports and newly released models that fit it.

Email me new reports

Reported configurations

Individual observations from this page, not expected speeds or a ranking. Select a model to open its report, with the source and offloading conditions. Reported context may be a limit rather than actual input length.

Model / hardware	Quant / engine	Reported context	Generation
Qwen3.6 35B (3B active) M5 32GB	Not reported oMLX	Not reported	40.0 tokens/s
Qwen3 30B (3B active) M5 32GB	4-bit MLX	Not reported	57.0 tokens/s
DeepSeek V4 Flash 284B (13B active) M5 32GB	4bit Engine not reported	Not reported	1.0 tokens/s

Frequently asked

How fast is the M5 32GB for local LLMs?

On the M5 32GB, Qwen3 30B · 3B active at 4-bit, the model most run on it, generates a median of 57.0 tokens per second from one plain run (community reports, as of 29 Oct 2026).