Alibaba Qwen legacy fast & cost-effective LLM (early 2025, Qwen2.5→Qwen3), 1M token context, MoE architecture. Focus on ultra-low latency + rock-bottom pricing. Ideal for high-concurrency chat, long-document summary, simple QA.