Reasoning
How to deploy Qwen3-Next-80B-A3B-Thinking on a GPU cloud
A 81.3B (MoE) model tuned to reason step by step before answering. Full specs, license and use cases.
Qwen3-Next-80B-A3B-Thinking size and hardware requirements
81.3B
Total parameters
~3B active per token (mixture-of-experts; see total parameters above)
Active parameters
BF16
Published precision
181.8 GB
Min VRAM (native)
| Precision | Weight size on disk | Required VRAM | Cheapest live fit | GPUs needed | Est. $/hr (full fit) |
|---|---|---|---|---|---|
| BF16 | 151.5 GB | 181.8 GB | RTX 3090 | 8 | $1.18/hr |
| FP8 (quantized) | 75.7 GB | 90.9 GB | RTX PRO 6000 | 1 | $1.64/hr |
| INT4 (quantized) | 37.9 GB | 45.4 GB | RTX A6000 | 1 | $0.330/hr |
How to run Qwen3-Next-80B-A3B-Thinking
Run Qwen3-Next-80B-A3B-Thinking with vLLM
From Qwen/Qwen3-Next-80B-A3B-Thinking's own deployment docs.
vllm serve Qwen/Qwen3-Next-80B-A3B-Thinking --port 8000 --tensor-parallel-size 4 --max-model-len 262144 --reasoning-parser deepseek_r1Source: https://huggingface.co/Qwen/Qwen3-Next-80B-A3B-Thinking/raw/main/README.md
Run Qwen3-Next-80B-A3B-Thinking with Ollama
Verified against Ollama's own library listing.
ollama run qwen3-next:80bRun Qwen3-Next-80B-A3B-Thinking with GGUF quantizations
Prebuilt GGUF weights published at skymizer/Qwen3-Next-80B-A3B-Thinking-GGUF. Run with llama.cpp's llama-server or load the repo directly in LM Studio.
llama-server -hf skymizer/Qwen3-Next-80B-A3B-Thinking-GGUFSource: https://huggingface.co/skymizer/Qwen3-Next-80B-A3B-Thinking-GGUF
Deploy Qwen3-Next-80B-A3B-Thinking on Aquanode
Aquanode has no one-click deploy template for Qwen3-Next-80B-A3B-Thinking; you install the inference engine yourself with the commands below. Aquanode sells GPU pods billed per second, not a hosted inference API.
- Launch a bare GPU pod sized to the requirement above (8× RTX 3090 or larger).
- Open a terminal on the pod, or save one of the commands above as a startup script so it runs automatically the first time the pod boots.
- Run the command and connect to the resulting endpoint.
Launch a GPU pod