Qwen3 30B

Qwen3 30B loads in 22 GB at Q4_K_M and is comfortable from 48 GB of memory, Macs included from the MacBook Pro M5 Pro 48GB. Below: the full memory math, the cheapest card that runs it well, and the fastest machine we track.

PARAMETERS
30B (3B active)
FORMAT
Q4_K_M
MIN MEMORY
32 GB
BEST FOR
Quality, Coding

Memory math

Weights (Q4_K_M)
22 GB
+ KV cache (16k)
~4.0 GB
Total at 16k
~26.0 GB
Comfortable from
48 GB

KV = fp16 estimate (q8_0 cache roughly halves it). "Comfortable" = weights + KV within the engine's tiered budget (~70-85% of memory).

Hardware snapshot

Cheapest GPU
~26 tok/s est. — ~$3,847 used (as of 2026-07-31)
Fastest
~125 tok/s est.
Runs on a Mac?
~42 tok/s est.

Go deeper

Every Qwen3 30B quant by real GGUF file size

More Qwen models

Qwen3.7-PlusQwen2.5 0.5B InstructQwen3.5 0.8B InstructQwen2.5 1.5B InstructQwen3.5 2B InstructQwen2.5 3B InstructQwen3.5 4B InstructQwen2.5 7B InstructQwen2.5 Coder 7BQwen3 8BQwen3.5 9B InstructQwen2.5 14B InstructQwen2.5 Coder 14BQwen3 14BQwen3.5 27B InstructQwen3.6 27BQwen3.8 27BQwen3 30B (Q8)Qwen3.5 35B-A3B InstructQwen3.6 35B-A3BQwen3-Next 80B-A3BQwen3.5 122B-A10B InstructQwen3.8-Flash-NextQwen3 235B A22B

Frequently asked questions

How much memory does Qwen3 30B need?

22 GB for the Q4_K_M weights, plus ~4.0 GB of KV-cache at 16k context — about 26.0 GB total. Comfortable from 48 GB of VRAM or unified memory.

Does Qwen3 30B run on a Mac?

Yes — from the MacBook Pro M5 Pro 48GB (~42 tok/s est.). Unified memory means the RAM budget is the only limit.

What is the cheapest GPU for Qwen3 30B?

The AMD Ryzen AI Max+ 395 (Strix Halo) is the cheapest tracked card that runs Qwen3 30B comfortably — ~26 tok/s est. at ~$3,847 used (as of 2026-07-31).

Cite this page

ModelFit: Qwen3 30B — specs, memory math and hardware verdicts.
https://modelfit.io/models/qwen3-30b/ (dataset updated 2026-09-03, CC BY 4.0).