Alibaba Group Holding Ltd.’s latest large language model, Qwen 3.6-Plus, ranked
second globally on Code Arena, a coding-focused benchmark under LMArena,
becoming the highest-ranked Chinese model. The model trailed Anthropic’s
Claude-Opus-4.6-Thinking (1540) but outscored offerings from OpenAI and Google,
including GPT-5.0-High (1448) and Gemini 3.1 Pro Preview (1440). Released April
2, Qwen 3.6-Plus features native multimodal reasoning and strong coding and
agent capabilities, with further models, including a Qwen 3.6-Max version,
expected soon.