On January 28, Chinese tech giant Alibaba launched the Qwen 2.5 AI model, which, according to the company, outperforms DeepSeek-V3.

Qwen 2.5-Max surpasses GPT-4o, DeepSeek-V3, and Llama-3.1-405B across multiple benchmarks, according to Alibaba’s cloud division.

The model was trained on over 20 trillion tokens and is available via API and Qwen Chat bot.

Two days before the release, Alibaba introduced Qwen2.5-VL, a new model family with multimodal capabilities. These models can process text, images, and videos, count objects, and even interact with computers in a manner similar to OpenAI’s AI agents.

According to benchmarks, Qwen2.5-VL outperforms GPT-4o, Claude 3.5 Sonnet, and Gemini 2.0 Flash in several key areas.

Hugging Face’s CTO, Philipp Schmid, demonstrated how Qwen2.5-VL successfully booked a flight on Booking.com.

The recent release of DeepSeek-R1 shook up the AI industry, triggering stock market dips and sparking debates over China’s rising dominance in artificial intelligence.

OpenAI’s CEO, Sam Altman, has since vowed to surpass DeepSeek.