A dense AI model with 32B parameters, excelling in coding, math, and local deployment. Compact, efficient, and powerful ...

Companies can freely deploy Light-R1-32B in commercial products, maintaining full control over their innovations.

scoring 96.7% on the American Invitational Mathematics Examination (AIME), surpassing its predecessor, o1. Yet, since Alibaba’s Qwen 2.5 launched, it has been a top competitor of both DeepSeek ...

This remarkable outcome underscores the effectiveness of RL when applied to robust foundation models pre-trained on extensive ...

The model, which was released on open source outperformed DeepSeek's R1, which boasts 671 billion parameters, in areas such ...

Hong-Kong listed shares (HK:9988) jumped over 8% today, after the company launched its new open-source AI (artificial ...

Qwen 2.5 VL, a new open source AI vision model promises to transform the way we approach visual understanding and automation. At its core, Qwen 2.5 VL isn’t just another AI tool—it’s a ...

Per the Qwen team’s benchmarking, the best Qwen2.5-VL model beats OpenAI’s GPT-4o, Anthropic’s Claude 3.5 Sonnet, and Google’s Gemini 2.0 Flash on a range of video understanding ...

Qwen’s latest reasoning model is part of an AI system replicating how humans think, make decisions, and solve problems. The Qwen team said it would open-source its coming QwQ-Max model and the ...

Alibaba has released its open-source QwQ-32B AI model, which is small enough to run on consumer-grade hardware, but claims to be on par with DeepSeek R1.

It ranks first in math and coding, and second in hard prompts ... They can also experience the model's capability on the Qwen Chat platform. Over the past year, the global cloud leader has ...