Alibaba Group has announced its new flagship AI model, Qwen4-Max, featuring a Mixture of Experts architecture and an impressive 2.4 trillion parameters. This advanced model boasts a context window of one million tokens and is designed for a range of applications including office productivity, creative endeavors, software development, and scientific research. Alibaba highlights significant cost reductions for task completion, with reports indicating a decrease of approximately 67% compared to its predecessor, Qwen3.8-Max. AI
IMPACT Sets a new benchmark for parameter count and context window size, potentially driving competition in large model development and cost-efficiency.
RANK_REASON Frontier-lab model release with system card. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
Read on Mastodon — mastodon.social →
- Alibaba Cloud
- Alibaba Group
- Artificial Analysis
- mixture of experts
- Qwen
- Qwen3.8-Max
- Qwen4-Max
- Qwen-Live-Delivery-Plugin
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →