Alibaba Launches Qwen3.6 with Advanced Agentic Coding, Multimodal AI, and Support for 201 Languages
Summary
Alibaba launches Qwen3.6, a powerful new AI model boasting advanced agentic coding, multimodal capabilities, and support for 201 languages, now available on Hugging Face Hub under an open Apache 2.0 license.
Key Points
- Alibaba's Qwen team releases Qwen3.6, the latest in their large language model series, with a focus on agentic coding, featuring improved front-end workflows, repository-level reasoning, and a new thinking preservation feature that retains context across conversation history.
- Qwen3.5, the foundation of Qwen3.6, introduces major upgrades including unified vision-language training on trillions of multimodal tokens, a hybrid architecture combining Gated Delta Networks with sparse Mixture-of-Experts, reinforcement learning scaled across million-agent environments, and expanded support for 201 languages.
- The Qwen3.6-35B-A3B model is now available on Hugging Face Hub and ModelScope, supports deployment via SGLang and vLLM with OpenAI-compatible APIs, and can be fine-tuned using frameworks like UnSloth, Swift, and Llama-Factory under an Apache 2.0 license.