Alibaba Unveils ‘Thinking’ AI Model QwQ-32

Chinese firm Alibaba has introduced the QwQ-32, an AI model focused on reasoning. With 32 billion parameters, it is purportedly capable of competing with leading solutions from rivals like DeepSeek-R1, which boasts 671 billion parameters.

Today, we release QwQ-32B, our new reasoning model with only 32 billion parameters that rivals cutting-edge reasoning model, e.g., DeepSeek-R1.

Blog: https://t.co/zCgACNdodj
HF: https://t.co/pfjZygOiyQ
ModelScope: https://t.co/hcfOD8wSLa
Demo: https://t.co/DxWPzAg6g8
Qwen Chat:… pic.twitter.com/kfvbNgNucW

— Qwen (@Alibaba_Qwen) March 5, 2025

The company has integrated agent capabilities and critical thinking into the neural network. QwQ-32’s performance is comparable to models from DeepSeek and OpenAI across several benchmarks assessing mathematical abilities, programming skills, and general problem-solving capabilities.

The new neural network is available in Qwen Chat, where users can select various models, including the most powerful — Qwen2.5-Max.

Alibaba’s shares rose by 8% during the trading session on March 5.

In November 2024, Alibaba introduced the ‘reasoning-oriented’ AI model QwQ-32B-Preview.

Previously, the company launched a group of neural networks focused on mathematics called Qwen2-Math, which ‘outperform GPT-4o and Claude 3.5’ in this domain.