Overview
Qwen is alibaba’s family of large language models developed by Alibaba Cloud’s AI lab. Competes with Western frontier models in coding, agentic reasoning, and multimodal orchestration.
Recent Developments
-
2026-09-17: Released qwen-3-8-omni-flash — 1M-token multimodal agent model with qwen-mm-plugins; $0.15/M input tokens; 6-region API (2026-09-17-qwen-3-8-omni-flash)
-
2026-08-03: qwen3-8-max GA on Model Studio / qwenwork; first Max-class open-weight commitment (2026-08-03-alibaba-qwen38-max-alicloud-blog)
-
2026-07-19: Launches qwen-38 / Qwen3.8-Max-Preview (2.4T); preview on Token Plan / qoder / QoderWork; open weights promised soon (alibaba-qwen-38-24t-near-frontier, 2026-07-19-alibaba-qwen-38-scmp)
-
2026-07-15: Disabling humanlike interactive agents (July 10) and full agent functions per anthropomorphic-ai-regulation (2026-07-06-china-bans-anthropomorphic-ai-personas)
-
2026-06: anthropic alleged Qwen lab-linked operators conducted industrial-scale distillation attack on claude (2026-06-25-anthropic-alibaba-distillation-campaign)
Related
- alibaba
- model-distillation
- anthropic
- anthropomorphic-ai-regulation
- qwen-38
- frontier-models
- kimi-k3
- mixture-of-experts
- qwen3-8-27b
- alibaba-cloud-model-studio