Alibaba's Latest Innovation in AI

Alibaba Group Holding Limited has taken a significant step forward in artificial intelligence with the launch of its Qwen 3 series of large language models (LLMs). This new series promises enhanced performance, scalability, and efficiency, marking a notable advancement in the field.

Competitive Edge in Benchmark Evaluations

The flagship model, Qwen3-235B-A22B, has demonstrated competitive results across various benchmarks, including coding, math, and general capabilities. It stands toe-to-toe with other leading models such as DeepSeek-R1, OpenAI's o1 and o3-mini, xAI's Grok-3, and Alphabet Inc.'s Gemini-2.5-Pro.

Introducing Hybrid Thinking Modes

What sets the Qwen 3 models apart is their innovative problem-solving approach, featuring two distinct modes: Thinking and Non-Thinking. Alibaba claims this hybrid method significantly improves the model's ability to manage thinking budget control efficiently and stably.