Qwen 3.8 27B shows a 17GB open-weight general purpose model can have long context, effective tool calling, strong vision ability, and competent code generation (Simon Willison/Simon Willison's Weblog)
Friday's big release was Qwen 3.8 27B, an Apache 2 licensed 27B parameter vision-capable LLM from Alibaba's Qwen research lab.
Alibaba's Qwen research lab recently unveiled Qwen 3.8 27B, a vision-capable 27-billion parameter language model. This model impressed users with its long context handling, effective tool calling, strong visual abilities, and proficient code generation. On a 128GB MacBook Pro, the model ran using LM Studio and a 17GB quantized build.
The default reasoning effort setting proved inefficient, but manually adjusting the setting revealed the model's full potential. Users were able to generate intricate SVG images with the help of Qwen 3.8 27B and saw the model accurately drawing bounding boxes around objects in photographs. A sample HTML page was provided to allow users to input image URLs and JSON data, enabling the display of scaled bounding boxes over the uploaded images.
Written by urgent.news from Techmeme's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
