Qwen3-Max

Abstract

Following the release of the Qwen3-2507 series, we are thrilled to introduce Qwen3-Max — our largest and most capable model to date. The preview version of Qwen3-Max-Instruct currently ranks third on the Text Arena leaderboard, surpassing GPT-5-Chat. The official release further enhances performance in coding and agent capabilities, achieving state-of-the-art results across a comprehensive suite of benchmarks — including knowledge, reasoning, coding, instruction following, human preference alignment, agent tasks, and multilingual understanding. We invite you to try Qwen3-Max-Instruct via its API on Alibaba Cloud or explore it directly on Qwen Chat. Meanwhile, Qwen3-Max-Thinking — still under active training — is already demonstrating remarkable potential. When augmented with tool usage and scaled test-time compute, the Thinking variant has achieved 100% on challenging reasoning benchmarks such as AIME 25 and HMMT. We look forward to releasing it publicly in the near future.