Paper summary
Paper summary: Qwen3 Technical Report
Qwen3 merges a "thinking" mode for multi-step reasoning and a fast non-thinking mode in one model family (0.6B to 235B, dense and MoE), adds a thinking budget, and expands to 119 languages.
- One model can either think step by step or answer directly, switched by the user or the chat template.
- A "thinking budget" lets users trade latency for reasoning depth at inference time.