Qwen3-Next-80B-A3B-Instruct

Alibaba · 81.3B (active 3B) · MoE · ctx 256K · apache-2.0 · HuggingFace

Qwen3-Next-80B-A3B-Instruct (Alibaba) is a 81.3B parameters MoE (3B active per token) open-weight model tracked on AI Hashrate. At 8K context we estimate roughly 46.32 GB VRAM for Q4 (weights 44.72 GB + KV 0.6 GB + 1 GB overhead) and 164.2 GB for FP16. About 11 GPUs/accelerators in our catalog fully fit this model at Q4 under that context assumption. Fastest Q4 configs: B200 192GB SXM ≈ 1697.0 tok/s (estimated); MI300X 192GB ≈ 1124.2 tok/s (estimated); H200 141GB SXM5 ≈ 1018.2 tok/s (estimated); H100 80GB SXM5 ≈ 710.6 tok/s (estimated); Gaudi 3 128GB ≈ 694.9 tok/s (estimated). Use the table below for VRAM fit and tok/s per dollar (list MSRP). Estimates follow memory-bandwidth math; measured rows override when present. See methodology for details. Methodology.

56 hardware configs — sorted by tok/s. Default fit context 8K.