Key Points

• 1Corroborated across 38 evidence documents
• 2I had Qwen build a Qwen-powered app — and sat in the reviewer's chair
• 3I stopped using Qwen and Gemma after finding a local LLM that actually thinks before answering
• 4Runapi-qwen-image added to PyPI
• 520B Looping model (paper) matches or beats Qwen3 Coder 30B at 10% of pre-training tokens
• 6Qwen3 embedding and reranking models for retrieval are now available in Amazon SageMaker JumpStart