Qwen3 Next 80B A3B Thinking

Qwen3-Next employs a highly sparse MoE design: 80 billion total parameters, but only approximately 3 billion are activated per inference step. Experiments show that, with global load balancing, increasing the total number of expert parameters while keeping the number of activated experts fixed steadily reduces training loss. Compared to Qwen3’s MoE (128 total experts, 8 routed), Qwen3-Next expands to 512 total experts, combining 10 routed experts + 1 shared expert — maximizing resource usage without compromising performance. The Qwen3-Next-80B-A3B-Thinking excels at complex reasoning tasks — outperforming higher-cost models like Qwen3-30B-A3B-Thinking-2507 and Qwen3-32B-Thinking, outperforming the closed-source Gemini-2.5-Flash-Thinking on multiple benchmarks, and approaching the performance of our top-tier model Qwen3-235B-A22B-Thinking-2507.