Alibaba Launches Its Largest AI Model Yet as China Races to Catch DeepSeek

Alibaba has officially entered the next phase of the AI arms race with the launch of Qwen3.8-Max, its largest and most capable artificial intelligence model to date.

Announced on August 3, the model boasts 2.4 trillion total parameters and utilizes a "mixture-of-experts" design, activating only 95 billion parameters per query to reduce costs and response delays.

The release comes just weeks after rival Moonshot AI unveiled its Kimi K3 model with 2.8 trillion parameters, signaling that Chinese tech giants are locked in a fierce battle to build more powerful yet cost-efficient systems. Investors took note, sending Alibaba's Hong Kong-listed shares up 7% on the day of the announcement.

The model is designed for multimodal understanding, capable of processing text, images, and video with a context window of up to 1 million tokens — enough to ingest entire legal files, large software codebases, or hundreds of pages of documents.

Alibaba claims the model autonomously completed a software engineering project over 16 days without human intervention, accumulating 265 commits and 151 issues.

On crowdsourced platform Arena.AI, Qwen3.8-Max immediately became the highest-ranking Chinese text model, ranking second globally for visual analysis — trailing only Anthropic's Claude Fable 5.

"Chinese tech companies — a huge force in open-weight AI models globally — are locked in a fierce and fast-moving battle to build more powerful systems without making them prohibitively expensive to run." – Reuters analysis of China's AI race

Unlike closed-source models from OpenAI, Anthropic, and Google, which do not publish parameter counts, Chinese AI developers are embracing open-weight transparency to gain traction among global developers.

Alibaba plans to release Qwen3.8-Max's weights for public download next week, marking the first time the company has open-sourced a model at this scale. This strategy is already paying off: Alibaba's Qwen family has surpassed Meta's Llama in global cumulative downloads on HuggingFace.

As parameter counts become just one measure of competitiveness, Chinese AI labs are increasingly focusing on practical capabilities such as coding, reasoning, and enterprise usability.

The launch of Qwen3.8-Max, coupled with Alibaba's $53 billion commitment to cloud and AI infrastructure over three years, signals that China's AI industry is moving beyond the pursuit of scale toward building systems that developers and businesses are more likely to adopt.