qwen36

Qwen3.6 is a next-generation large language model series optimized for agentic coding and repository-scale reasoning.

Alibaba's Qwen3.6 series introduces a high-performance architecture designed for autonomous developer workflows and complex problem-solving. The flagship Qwen3.6-Plus model features a 1 million token context window and native multimodal reasoning (processing text, images, and video) to handle entire codebases or dense technical documentation without chunking. For open-source deployment, the Qwen3.6-35B-A3B variant utilizes a sparse Mixture-of-Experts (MoE) design: it leverages 35 billion total parameters but activates only 3 billion per token for high-throughput inference. This release prioritizes thinking preservation (retaining reasoning context across conversation turns) and achieves top-tier results on benchmarks like SWE-bench Verified and Terminal-Bench 2.0.

https://qwen.ai