Alibaba Cloud (Qwen) - Mixture of experts - verified
Run Qwen3-Coder-Next locally
Qwen3-Coder-Next is a 79.7B mixture-of-experts (3B active) model from Alibaba Cloud (Qwen) with a 262K-token context window, released under Apache License 2.0 (commercial use is allowed); the smallest listed download is 44.8 GB, the 4bit MLX build from mlx-community/Qwen3-Coder-Next-4bit.
Official weights: Qwen/Qwen3-Coder-Next
- Total params
- 79.7B
- Active params
- 3B
- Context
- 262K
- Smallest download
- 44.8 GB
Fits in
Smallest download at most 85% of memoryDownloads
The official weights and every quantization listed for this model, with file sizes from Hugging Face.
| Download | Format | Size | Repo |
|---|---|---|---|
Official weights | BF16 | 159.4 GB | Qwen/Qwen3-Coder-Next |
Q4_K_M | GGUF | 48.4 GB | Qwen/Qwen3-Coder-Next-GGUF |
Q8_0 | GGUF | 84.8 GB | Qwen/Qwen3-Coder-Next-GGUF |
4bitSmallest listed | MLX | 44.8 GB | mlx-community/Qwen3-Coder-Next-4bit |
GB is decimal: 1 GB = 10^9 bytes. Add memory for the context and the operating system on top of the file size.
Run it
Copied verbatim from the page linked under each command. Server commands from model cards often assume a multi-GPU node; check the flags before you run them.
llama.cpp
./llama-cli -m ./Qwen3-Coder-Next-Q5_K_M/Qwen3-Coder-Next-00001-of-00004.gguf --jinja -ngl 99 -fa on -sm row --temp 1.0 --top-k 40 --top-p 0.95 --min-p 0 -c 40960 -n 32768 --no-context-shiftvLLM
vllm serve Qwen/Qwen3-Coder-Next --port 8000 --tensor-parallel-size 2 --enable-auto-tool-choice --tool-call-parser qwen3_coderSGLang
python -m sglang.launch_server --model Qwen/Qwen3-Coder-Next --port 30000 --tp-size 2 --tool-call-parser qwen3_coderNot sure which runtime to use? Compare local LLM runtimes by platform, API and features.
License
- License
- Apache License 2.0
apache-2.0 - Commercial use
- Commercial use: YesCommercial use is allowed.
- Gated download
- No. The files download directly.
Worth knowing
Non-thinking mode only (no think blocks), and the card suggests cutting context to 32,768 tokens if you hit out-of-memory errors; needs vllm>=0.15.0 or sglang>=v0.5.8.
Released (Hugging Face repo creation date).
FAQ
How big is the Qwen3-Coder-Next download?
The official BF16 weights in Qwen/Qwen3-Coder-Next are 159.4 GB. The smallest listed download is 44.8 GB, the 4bit MLX build from mlx-community/Qwen3-Coder-Next-4bit. Plan for more memory than the file size: the context (KV cache) and the operating system need room too.
Can I use Qwen3-Coder-Next commercially?
Qwen3-Coder-Next is released under Apache License 2.0, and commercial use is allowed. The license text is at https://www.apache.org/licenses/LICENSE-2.0.
How much memory do I need for Qwen3-Coder-Next?
By this directory's rule of thumb it fits in 64 GB and up: the smallest listed download, 44.8 GB, is at most 85% of 64 GB, which leaves room for the context and the operating system.
How do I run Qwen3-Coder-Next with Ollama?
Run "ollama run qwen3-coder-next", as shown at https://ollama.com/library/qwen3-coder-next. The Ollama library lists the qwen3-coder-next tag at 52GB.
What is the context length of Qwen3-Coder-Next?
262,144 tokens. A longer context needs more memory for the KV cache, so you can run with a shorter one when memory is tight.
Sources
Last verified from the pages below.
- HF API (params, license, createdAt) - huggingface.co
- HF file tree - huggingface.co
- Model card (80B total, 3B activated) - huggingface.co
- Official GGUF card - huggingface.co
- Official GGUF tree - huggingface.co
- mlx-community 4bit tree - huggingface.co
- Ollama library - ollama.com
Newsletter
New open weights, every week
One email per week. New model releases, open-weight drops and what they mean for what you build.
Keep exploring
Best local AI models
Our curated picks by hardware tier, with benchmarks and the reasoning.
Model API pricing and specs
First-party prices per 1M tokens when you would rather call an API.
AI benchmark leaderboards
Sourced scores, for when you need to compare quality.
Setup guides
Point your coding tools at the model you picked.