QriticalBeta
Open Model Catalogue
Alibaba Qwen
Alibaba Qwen
Text generationCoding

Qwen3-Next 80B MoE

Qwen/Qwen3-Next-80B-A3B-Instruct

Ultra-fast sparse MoE — 2 × H200

Deploy Qwen3-Next 80B MoE nowView on HuggingFace
Parameters
80B MoE · 3B active
Min VRAM
≈ 190 GB
Hardware
3 × 80 GB GPU
OpenAI-compatible API with a per-workload key — drop-in for existing SDKs.
Dedicated GPUs — your model, your instance, no shared queues.
Billed by the second with auto-shutdown — never pay for idle time you didn't plan.