Junyang Lin Leaves Qwen Team

🔑 Enhanced Key Takeaways

• •Junyang Lin announced his departure from Alibaba's Qwen team on X (@JustinLin610) with the post 'me stepping down. bye my beloved qwen,' which received hundreds of retweets and thousands of likes[1].
• •Another Qwen researcher, Chen Chang, indicated Lin's exit was involuntary, stating they launched Qwen3.5 small model together the night before and expressing heartbreak over Qwen without him[1].
• •Lin joined Alibaba in 2019 as a Senior Algorithm Engineer in NLP and multimodal research, rising to Senior Staff Algorithm Engineer and Qwen tech lead from 2023[1].
• •Lin also developed earlier models M6 and Tongyi Qianwen, establishing him as a key figure in Alibaba's AI efforts[3][4].

🛠️ Technical Deep Dive

• •Qwen3.5 series includes open-weight Mixture of Experts model Qwen3.5-397B-A17B, emphasizing serving efficiency[5].
• •Proprietary Qwen3.5-Plus supports 1M token context length, native 256K tokens, search, and code interpreter via hosted API[5].
• •Qwen3.5 models are multimodal, supporting vision input for native multimodal agents[5].

🔮 Future ImplicationsAI analysis grounded in cited sources

⏳ Timeline