- Updated: March 4, 2026
- 2 min read
Qwen 3.5 Model Family Launched Amid Leadership Changes at Alibaba
Alibaba’s AI research division has unveiled the powerful Qwen 3.5 model family, a suite of large language models that promise state‑of‑the‑art performance across a range of tasks. The release comes at a turbulent time for the team, as several senior engineers have recently resigned, raising questions about the long‑term direction of the project.
The Qwen 3.5 series builds on the success of its predecessor, offering larger context windows, improved multilingual capabilities, and more efficient inference. According to the official release, the models achieve competitive scores on benchmarks such as MMLU, CEVAL, and HumanEval, positioning them alongside leading open‑source alternatives.
While the technical advancements are impressive, the leadership shake‑up has sparked concern within the community. Simon Willison highlighted the departures in a recent blog post, noting that the loss of key talent could affect future development cycles and the sustainability of the open‑source roadmap.
Despite the uncertainty, Alibaba remains committed to supporting the Qwen ecosystem. The company has pledged continued investment in research, community engagement, and open‑source contributions. For developers and enterprises looking to experiment with cutting‑edge LLMs, Qwen 3.5 represents a compelling option that balances performance with accessibility.
Read the full analysis by Simon Willison here. For more AI news and insights, explore our internal resources:
Andrii Bidochko
CTO UBOS
Andrii Bidochko is an AI entrepreneur and researcher focused on AI agents, reinforcement learning, and autonomous systems. He writes about the technologies shaping the future of machine intelligence, from frontier models and agent architectures to real-world AI applications.