Related Links
7HomeSec-Bench
HomeSec-Bench results show Qwen3.5-9B scores 93.8% on real security AI tests, within 4 points of GPT-5.4, when running entirely on Apple Silicon.
Something is afoot in the land of Qwen
Simon Willison is behind on writing about Qwen 3.5, a family of open-weight models released by Alibaba's Qwen team. The models have garnered attention in the AI community due to their capabilities.
Qwen3.5 Fine-tuning Guide
The Unsloth documentation provides a guide for fine-tuning Qwen3.5 large language models (LLMs). The guide likely details the tools and techniques for adapting the Qwen3.5 model for specific tasks or datasets using the Unsloth framework.
Alibaba's Qwen tech lead steps down after major AI push
Junyang Lin has stepped down as the tech lead of Alibaba's Qwen large language model team after a major AI push. The move triggered reactions within the Qwen team at Alibaba.
Alibaba's new open source Qwen3.5-Medium models offer Sonnet 4.5 performance on local computers
Alibaba has released new open-source Qwen3.5-Medium language models that reportedly offer performance comparable to Google's Sonnet 4.5 on local computers. This performance leap is attributed to near-lossless accuracy under 4-bit weight and KV cache quantization, enabling developers to process large datasets without needing server-grade infrastructure.
Unsung - Waxy.org
This blog post from Waxy.org provides a collection of various news, cultural observations, and internet ephemera from the past week. It covers a wide range of topics, including AI, sports, media, business, and various cultural trends.
Qwen
The linked page describes Qwen Chat, an AI assistant with capabilities spanning chatbot functionality, image and video understanding, image generation, document processing, web search integration, tool utilization, and artifact creation.