Alibaba launched Qwen3.8-Max, its largest model to date. It has 2.4 trillion parameters. It supports a 1 million token context window. It ranks fifth in Text Arena and second in Vision Arena. It uses a sparse mixture-of-experts design that activates only 95 billion parameters. It is available now via Alibaba Cloud Model Studio API. Model weights will be released next week. It can also be used on QwenWork. It autonomously built a self-evolving agent framework over 16 days. It outperformed humans in the WWW2025 Multimodal Dialogue Intent Recognition Challenge. It can ingest hundred-page documents or 100-hour livestreams into searchable knowledge bases. It can edit raw footage into vlogs or turn 2D plans into 3D visualizations. It reconstructed applications from scratch using only interaction and feedback.
What changed
Qwen 3.5 was the prior flagship model.
What it unlocks
Running a 2.4 trillion parameter model with only 95 billion activated at a time.
- 2.4 trillion parameters
- 1 million token context window
- 95 billion activated parameters
- 16-day autonomous project
What you need to act on it
- access via Alibaba Cloud Model Studio API
- model weights release scheduled for next week
Sources