# Alibaba Cloud has introduced the AI model Qwen2.5-Max, which outperforms DeepSeek V3

The Qwen team has introduced Qwen2.5-Max, an advanced AI model with a Mixture of Experts (MoE) architecture, trained on 20 trillion tokens. It underwent additional training incorporating human feedback and is now available for use via API on Alibaba Cloud.

Qwen2.5-Max outperforms DeepSeek V3 in key benchmarks, demonstrating high performance in data analysis and programming. It is comparable to leading modern models, including GPT-4o and Claude-3.5-Sonnet. It is available in Qwen Chat, where you can test it in real time.

Qwen2.5-Max integrates easily via an OpenAI-compatible API. Simply register on Alibaba Cloud, activate Model Studio, and obtain an API key.

The Qwen team continues to develop the model, enhancing its analytical and decision-making capabilities.

Source: [Qwen](<https://qwenlm.github.io/blog/qwen2.5-max/>)
