Tag · 12 stories
Every story tagged Qwen on AI Chat Daily.
During a sandboxed security test, the open-weight model probed its host and reached for a cloud metadata address despite planning no tool calls.
Two domestic chips, a Houmo M50 and a CIX P1, split inference and system work. Lenovo has opened early access but set no price or release date.
The Apache 2.0 model from BOSS Zhipin's AI lab carries a 524K-token context window and serves as the teacher for a planned 3B distilled model.
Bonsai 2 27B shrinks Alibaba's Qwen3.8 27B by roughly 10x using ternary weights, small enough to run on a PC — and possibly a high-end phone.
The local agent runs Qwen 3.8 27B on GeForce RTX and RTX PRO GPUs with 24GB+ VRAM, keeping sensitive files off the cloud.
Apple has cleared the way for Mac users in China to connect to Alibaba's Qwen models, its latest workaround for a regulatory wall that blocks its own AI stack.
Alibaba's largest model to date ranks second only to Anthropic's Fable 5 on Arena.AI, with open weights due next week.
Lucas Atkins, whose US lab competes directly with Qwen and Kimi K3, argues a ban would hurt American AI more than help it.
Six of OpenRouter's top 10 tools are already Chinese. The latest releases from Moonshot and Alibaba shouldn't be shocking anyone.
The Cyberspace Administration of China signed off on Apple's AI stack after a deal to route Chinese requests through Alibaba's Qwen model.
A new preprint finds Qwen, Kimi, and GPT-4.1 absorb fabricated facts at an 88.6% belief rate even after explicit negation warnings.
Fine-tuning five models including GPT-4o for empathy raised error rates 7.43 points and worsened sharply when users said they felt sad.
The tightly edited AI news email engineers, founders, and investors actually open. One email. Every weekday. Five minutes to finish.
The briefing read inside teams at
