Alibaba (09988) has released its new Qwen3 series of AI models, claiming they can rival the best offerings from Google and OpenAI and even outperform competitors in certain scenarios. The series includes two MoE models and six Dense models, with parameters ranging from 600 million to 235 billion.
The flagship model, Qwen3-235B-A22B, delivers competitive results in coding, mathematics, and general capabilities when compared to other leading models such as DeepSeek-R1, OpenAI's o1 and o3-mini, Elon Musk's Grok 3, and Google Gemini 2.5 Pro. The smaller MoE model, Qwen3-30B-A3B, boasts an activation parameter size that is only 10% of QwQ-32B, yet it shows superior performance. Even the small Qwen3-4B model rivals the performance of Qwen2.5-72B-Instruct.
The six Dense models, including Qwen3-32B, Qwen3-14B, Qwen3-8B, Qwen3-4B, Qwen3-1.7B, and Qwen3-0.6B, have been made open source under the Apache 2.0 license. Qwen3 supports two thought modes: a deep reasoning mode for complex problems and a non-thinking mode for rapid responses to simpler queries. The series demonstrates enhanced reasoning abilities, achieving state-of-the-art levels in mathematics, code, and logical reasoning evaluations.
