On March 3, 2025, Tongyi Lingma officially released its latest inference model, Qwen2.5-Max, which provides developers with powerful programming and mathematical capabilities. The launch of Qwen2.5-Max marks another major breakthrough in artificial intelligence technology in the fields of programming and mathematics. The model is based on pre-training data of over 20 trillion tokens and combined with a carefully designed post-training scheme, bringing it to new heights in performance.
Qwen2.5-Max performed particularly well in several benchmarks. For example, in tests such as Arena-Hard, LiveBench, LiveCodeBench and GPQA-Diamond, Qwen2.5-Max scores surpass other leading models in the industry, including DeepSeek V3, GPT-4o, and Claude-3.5-Sonnet. In addition, in MMLU-Pro and other evaluations, Qwen2.5-Max also showed extremely competitive results, further proving its outstanding ability in complex tasks.
In the comparison of the base model, Qwen2.5-Max was comprehensively compared with DeepSeek V3, Llama-3.1-405B and Qwen2.5-72B. The results show that Tongyi Qianwen's base model shows significant advantages in most benchmarks, especially when dealing with complex programming and mathematical problems, Qwen2.5-Max performs particularly well.
It is worth noting that in the latest blind test list of large models released by the tripartite benchmark test platform Chatbot Arena, Qwen2.5-Max surpassed models such as DeepSeek-V3, Open AI o1-mini and Claude-3.5-Sonnet, ranking seventh in the world with 1332 points, becoming the champion of China's non-inference big models. In terms of single-item abilities such as mathematics and programming, Qwen2.5-Max ranks first, and Hard prompts ranks second. Chatbot Arena official commented that Alibaba's Qwen2.5-Max performed strongly in many fields, especially in professional and technical fields such as programming, mathematics and hard prompts.
At present, Qwen2.5-Max has been integrated into Tongyi Lingcode, and users can experience its powerful programming capabilities by downloading Tongyi Lingcode plug-in. This integration not only provides developers with more efficient tools, but also opens up new possibilities for the application of artificial intelligence technology.