Qwen 3.5-9B Beats GPT-OSS-120B (13× Larger) on GPQA, MMLU-Pro, MMMU — Runs on Standard Laptops

VentureBeat·high signal

Alibaba's Qwen 3.5-9B (released March 1) outperforms OpenAI's GPT-OSS-120B on GPQA Diamond (81.7 vs. 71.5), MMLU-Pro (82.5 vs. 80.8), and multilingual MMMLU (81.2 vs. 78.2) — despite being 13× smaller. The model uses a hybrid Gated Delta Network + sparse MoE architecture with a 262K native context window extensible to 1M, and runs efficiently on consumer laptops. All four models in the family (0.8B, 2B, 4B, 9B) are Apache 2.0 open-weight on Hugging Face.