Qwen 3-Coder

What Qwen 3-Coder is for

Qwen 3-Coder is trained for code. Alibaba released it in 2025-07 with 2 sizes (30B, 480B), 256K of context and a Apache 2.0 (30B), custom (480B) licence. Context length matters more here than for chat, because a coding assistant that cannot see the whole file is guessing. The catalogue records its strength as long-context code agent workflows, and it is the class of model an editor plugin such as Continue or Tabby points at.

Sizes and memory

The 30B needs about 18 GB of weights at Q4_K_M and the 480B about 280 GB, a spread of 262 GB across the family. The catalogue's default pick is 30B. Recommended VRAM adds roughly half again on top of the weights for the KV cache and the runtime.

Size	Weights at Q4_K_M	Recommended VRAM	Smallest card in our guide that holds it
30B	18 GB	27 GB	NVIDIA RTX 3090
480B	280 GB	420 GB	None in this guide

Hardware that runs Qwen 3-Coder

11 of the 19 cards in our hardware guide hold at least one size of Qwen 3-Coder at Q4_K_M. The cheapest that does is the NVIDIA RTX 3090 at about $700, running the 30B.

The largest size any of them holds is the 30B at 18 GB, on the NVIDIA RTX 3090 (24 GB at 936 GB/s). Bandwidth, not capacity, sets the speed once a size fits, so a card further down the list with faster memory will generate more tokens per second on the same weights than a larger, slower one.

What people use it for

The catalogue lists 3 recorded uses for Qwen 3-Coder: code agents, long-context code, refactoring. It is tagged code, code-agents, long-context, MoE.

Where Qwen 3-Coder sits in the Qwen family

Our catalogue holds 5 Qwen entries. Ordered by release date, Qwen 3-Coder (2025-07) is the 4th of them, and Qwen 3 VL (2025-10) is newer.

The Apache 2.0 (30B), custom (480B) licence puts no usage ceiling on it, which matters if the thing you are building has users. That is the practical difference between this and a vendor community licence.

Context window: 256K

256K tokens is roughly 192,000 words at the usual ratio, which is a small repository or a whole book. Two cautions come with a window this size: the KV cache at full length can rival the weights for memory, and models rarely use the far end of their advertised context as well as they use the near end. Treat it as headroom, not as a promise.

How to run Qwen 3-Coder locally

Our catalogue does not record registry tags, so look the current tag up in the Ollama library or on Hugging Face before pasting these. The commands below are the shape of the workflow, not a copy-and-paste recipe.

Option 1: Ollama (simplest)

# find the tag for Qwen 3-Coder at ollama.com/library
ollama run <tag>

Option 2: Mullama (production)

mullama pull <tag>
mullama run <tag>

Option 3: llama.cpp (CLI)

# download a GGUF from Hugging Face, searching for "Qwen 3-Coder GGUF"
./llama-cli -m model.Q4_K_M.gguf -p "Hello, AI!"

Option 4: Python with Mullama or llama-cpp-python

from mullama import Model, Context
model = Model.load("model.Q4_K_M.gguf", n_gpu_layers=99)
ctx = Context(model, n_ctx=4096)
print(ctx.generate("Hello, AI!", 256))

Sources

Model record from src/data/models.json (2026-06-29); card capacities from src/data/gpus.json (2026-06-29). Upstream: Curated from ollama.com/library + community benchmarks + paperswithcode + huggingface.