Qwen Max 2025-09-23

Qwen Max 2025-09-23 Description

Qwen Max 2025-09-23, developed by Alibaba, is a large-scale Mixture-of-Experts (MoE) language model optimized for chat, coding, and complex problem-solving tasks. It employs a Transformer-based MoE architecture with expert routing, activating a subset of parameters per token to balance performance and compute efficiency. Pre-trained on over 20 trillion tokens from diverse multilingual and domain-specific corpora, Qwen Max supports deep reasoning and advanced text generation.

Technical Specifications

• Architecture: Transformer-based Mixture-of-Experts (MoE) with selective expert activation per input token
• Total Parameters: Not explicitly specified but aligned with large-scale MoE designs
• Pre-training Data: More than 20 trillion tokens, including extensive English and Chinese textual data and various major world languages
• Context Window: 262K tokens native
• Language Support: Primarily English and Chinese
• Training Methods: Supervised Fine-Tuning (SFT) and Reinforcement Learning from Human Feedback (RLHF)
• Intended Use: Chat interactions, code generation, and complex question answering

Key Capabilities

• Efficient MoE architecture enabling strong performance with reduced computational cost
• Supports bilingual proficiency in English and Chinese for broad application scenarios
• Excels in coding tasks, problem-solving, and conversational AI, facilitating code generation and complex reasoning
• Employs advanced fine-tuning strategies to enhance generation quality and alignment with human preferences

Performance Metrics

Usage

Code Samples:

The model is available on the AI/ML API platform as "Qwen Max 2025-09-23" .

API Documentation:

Detailed API Documentation is available here.

Ethical Guidelines and Licensing

• Alibaba emphasizes transparency about Qwen Max's capabilities and limitations
• Encourages responsible and ethical use to avoid misuse or harmful content generation
• Governed by specific licensing terms provided by Alibaba Cloud, with users advised to review usage permissions and restrictions

Get Qwen Max 2025-09-23 API here.