Alibaba challenges OpenAI

The Qwen-3-Max-Preview model has more than a trillion parameters. Photo: Reuters .

Alibaba Group Holding has just released its largest artificial intelligence model to date, marking a new step forward in the global AI race. Called Qwen-3-Max-Preview, this is the first model from the Chinese technology group to surpass the one trillion parameter threshold, putting Alibaba in the ranks of “giants” such as OpenAI, Google DeepMind or Anthropic.

The new model, released on September 5, is now available on Alibaba Cloud and the major language model marketplace OpenRouter. Qwen-3-Max-Preview follows the Qwen3 series, which was first introduced in May with versions ranging from 600 million to 235 billion parameters.

In AI, parameters are considered the “intelligence” of a system, the variables that are adjusted during training. A larger number of parameters typically means more powerful capabilities, but also requires massive computing power. For comparison, OpenAI’s GPT-4.5, one of the largest models available today, is estimated to have 5-7 trillion parameters.

According to the announcement, Qwen-3-Max-Preview is a text processing model that outperforms Qwen3-235B-A22B-2507. In internal tests, Qwen-3-Max-Preview outperformed MoonShot AI's Kimi K2, a non-inference version of Claude Opus 4, and DeepSeek V3.1, which supports Anthropic's format. However, these benchmarks have not been included in the official technical report.

“Qwen-3-Max-Preview shows significant improvements in overall capabilities, from understanding Chinese and English text, performing complex instructions, handling subjective tasks, supporting multiple languages, and using tools. The expansion is still ongoing, and the official release will bring more surprises,” Alibaba said in a statement.

Over the years, Qwen has helped Alibaba stand out in the global open-source AI ecosystem. These models have reached over 20 million downloads and generated 100,000 derivative models on the Hugging Face platform. However, Qwen-3-Max-Preview is not open source. Users can only access it via Alibaba Cloud or OpenRouter, similar to Qwen2.5-Max.

On social media, Binyuan Hui, an AI engineer at Alibaba, revealed that a “reasoning” version of the model is still in development.

The cost of accessing Qwen-3-Max-Preview on Alibaba Cloud is also among the highest in the Qwen family. The starting price is $0.861 per million input tokens and $3.441 per million output tokens. Meanwhile, the Qwen3-235B-A22B-2507 version costs only $0.287 per million input tokens and $1.147 per million output tokens.

Source: https://znews.vn/alibaba-cong-bo-mo-hinh-thach-thuc-openai-post1583459.html