China's Alibaba releases new AI model that claims to solve problems better than rivals DeepSeek, OpenAI

ANI March 6, 2025 298 views

Alibaba has unveiled its latest AI model QwQ-32B, marking a significant breakthrough in artificial intelligence technology. The model claims to rival larger competitors like DeepSeek and OpenAI by delivering impressive performance with fewer computational resources. Built on the Qwen 2.5 platform, the AI can process multiple data types and demonstrate advanced reasoning capabilities. Alibaba's strategic approach focuses on integrating reinforcement learning and agent-based technologies to push the boundaries of AI innovation.

"Our research explores the scalability of Reinforcement Learning (RL) and its impact on enhancing the intelligence of large language models" - Alibaba AI Team
Washington DC, March 6: Chinese e-commerce and tech giant Alibaba on Thursday unveiled a new artificial intelligence (AI) model, which it claims can rival DeepSeek to solve complex problems, but is better because it uses far less data for its operations

Key Points

1

Alibaba's QwQ-32B delivers high performance with just 32 billion parameters

2

Model built on Qwen 2.5 handles text, images, and audio

3

Outperforms DeepSeek's R1 in mathematics and coding

4

Aims to advance towards Artificial General Intelligence

It said that its new compact reasoning model QwQ-32B is "comparable" to other larger cutting edge models such as OpenAI's o1-mini.

"Today, we release QwQ-32B, our new reasoning model with only 32 billion parameters, delivering performance comparable to other larger cutting edge models," a X post by the Alibaba Group founded by billionaire Jack Ma said .

Alibaba's AI model follows in the wake of another Chinese company DeepSeek, which had in January 2025 released its low-cost AI model that rivalled OpenAI's Chat GPT, a leader in the field of Artificial Intelligence.

Alibaba's new model is built on Qwen 2.5, its latest Artificial Intelligence language model, which is designed to handle text, images, and audio to analyse complex data, identify patterns and generate solutions, in ways that humans do.

The model, which was released on open source outperformed DeepSeek's R1, which boasts 671 billion parameters, in areas such as mathematics, coding and general problem-solving, according to a posting on the Qwen, the company's blog post on its AI models.

In layman's terms, there are three main types of AI models- supervised learning, unsupervised learning, and reinforcement learning.

"Our research explores the scalability of Reinforcement Learning (RL) and its impact on enhancing the intelligence of large language models," the team looking after Alibaba AI models said.

"Furthermore, we have integrated agent-related capabilities into the reasoning model, enabling it to think critically while utilizing tools and adapting its reasoning based on environmental feedback," they said.

"These advancements not only demonstrate the transformative potential of RL but also pave the way for further innovations in the pursuit of artificial general intelligence," they added.

The company said it is confident that in future it can by "combining stronger foundation models with RL powered by scaled computational resources" get closer to achieving Artificial General Intelligence (AGI).

"Additionally, we are actively exploring the integration of agents with RL to enable long-horizon reasoning, aiming to unlock greater intelligence with inference time scaling," it said.

Leave a Comment

Your email won't be published

Tags:
Alibaba AI Model Jack Ma QwQ-32B Artificial Intelligence DeepSeek OpenAI Machine Learning
You May Like!