Alibaba Cloud has introduced the AI model Qwen2.5-Max, which outperforms DeepSeek V3
Author - Mansur Ismagulov
The Tech AI
Get the key points in seconds - a short summary from The Tech AI.
The Qwen team has introduced the Qwen2.5-Max model, featuring an MoE architecture and trained on 20 trillion tokens. It outperforms DeepSeek V3 and is comparable to GPT-4o and Claude-3.5-Sonnet. It is available via the Alibaba Cloud API and in Qwen Chat.
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
The Tech on Telegram Follow The Tech on Telegram
Keep up with the latest technology and startup news.
Related reading
More from this author
News
zypl.ai begins operations in Japan and focuses on next-generation risk models
Mansur Ismagulov
News
Elon Musk's net worth has exceeded $700 billion for the first time
Mansur Ismagulov
Article Interviews
"Central Asian countries should build a joint ecosystem," — Bekhzod Ismoilov on the development of AI in the region
Mansur Ismagulov
Article Interviews
Bloom's Taxonomy and Engineering Levels: A Practical Growth Model
Mansur Ismagulov
Article Useful
How the IT industry in Azerbaijan developed in 2025: key events
Mansur Ismagulov
News
Tajikistan has become a new member of the regional innovation alliance Central Asian Innovation Hubs
Mansur Ismagulov