Only with CPU offload

Qwen3 Coder 30B-A3B at Q4_K_M needs 18.5 GB but only 18.4 GB is addressable, so about 1% of the layers would stream from system RAM at roughly 60 GB/s. Expect around 94 tokens per second — usable for batch work, painful for chat.

99% on GPU 8K context Q4_K_M · 17.1 GB Apache 2.0 Released Jul 2025

Agentic coding MoE with a 256K native window. Still the most-downloaded local code model.

Hardware requirements What hardware do I need to run Qwen3 Coder 30B-A3B? →

At Q4_K_M with an 8K context: a 24 GB GPU (RTX 4090) or a 32 GB Mac (M4). Minimum, recommended, high-quality and long-context tiers, with all 114 devices checked at every quantisation.

Fits instead: Qwen3 14B (10.2 GB) · Qwen3 8B (6.3 GB)

The VRAM budget

weights 17.1 GB
Weights 17.1 GB KV cache @ 8K 0.75 GB Runtime overhead 0.6 GB Over budget 0.1 GB past 18.4 GB

Quantisation ladder

QuantWeightsTotal @ 8KMax contextTok/sQualityFit
Q8_0 30.2 GB 31.5 GB — ~9.0 −0.1% ppl 13.1 GB over
Q6_K 23.3 GB 24.6 GB — ~17 −0.4% ppl 6.2 GB over
Q5_K_M 20.1 GB 21.5 GB — ~30 −0.8% ppl 3.1 GB over
Q4_K_M 17.1 GB 18.5 GB 6K ~94 −1.9% ppl 0.1 GB over
Q3_K_M 13.9 GB 15.2 GB 41K 124 −5.4% ppl Long context
Q2_K 11.9 GB 13.2 GB 62K 145 −15% ppl Long context

Quality is the published perplexity delta against f16 weights. Max context assumes an f16 KV cache; q8_0 roughly doubles it.

How to run it

terminal
$ llama-server \
    -hf Qwen/Qwen3-Coder-30B-A3B-Instruct:Q4_K_M \
    -c 8192 -ngl 47

The engine underneath most of the others. Every knob is exposed. More on llama.cpp.

01Download is 17.1 GB. Keep it on an SSD — a first load off a spinning disk takes minutes.
02Close anything else holding VRAM. A browser with hardware acceleration can sit on 1–2 GB.
03If it falls back to CPU silently, drop the context first, then step down a quantisation.
See all models for this rig Compare with another model

Related models on a Radeon RX 7900 XT

ModelTotalTok/s est.
Qwen3 30B-A3B
30.5B MoE · same family
Q4_K_M · 18.5 GB ~94 CPU offload
Qwen3 32B
32.8B · same family
Q4_K_M · 21.0 GB ~9.5 CPU offload
Qwen3 14B
14.8B · same family
Q4_K_M · 10.2 GB 58 Runs great
Qwen3 8B
8.19B · same family
Q4_K_M · 6.3 GB 105 Runs great
GLM-4.7-Flash 30B-A3B
31.2B MoE · similar size
Q4_K_M · 18.6 GB ~99 CPU offload
Muse Glimmer 30B
29.8B · similar size
Q4_K_M · 17.5 GB 29 Tight fit
Gemma 4 31B
31.3B · similar size
Q4_K_M · 20.2 GB ~12 CPU offload

Every model against the Radeon RX 7900 XT →

Qwen3 Coder 30B-A3B on other hardware

DeviceTotalTok/s est.
GeForce RTX 4060 Laptop
7.0 GB usable · 256 GB/s
18.5 GB ~10 CPU offload
GeForce RTX 3060 12 GB
10.6 GB usable · 360 GB/s
18.5 GB ~14 CPU offload
M4 · 16 GB
10.7 GB usable · 120 GB/s
18.5 GB — Won't fit
GeForce RTX 4090
22.4 GB usable · 1008 GB/s
18.5 GB 126 Runs great
Radeon RX 7900 XTX
22.4 GB usable · 960 GB/s
18.5 GB 120 Runs great
M4 · 32 GB
24.0 GB usable · 120 GB/s
18.5 GB 17 Runs great
M3 Max · 64 GB
48.0 GB usable · 400 GB/s
18.5 GB 56 Runs great

All 114 devices, by tier →