[AI]■ STORY TIMELINE

PRISMML SHRINKS ALIBABA'S AI MODEL TO FIT SMARTPHONES

PrismML has compressed Alibaba's Qwen 3.8 27B model to 5.9 GB while maintaining 98.2% of its benchmark performance, making large language models feasible for mobile devices.

1 SOURCEFIRST SEEN SEP 18, 03:05 AM► READ THE ARTICLE

Techmeme+0m

PrismML releases Bonsai 2 27B, which compresses Alibaba's Qwen3.8 27B to 5.9 GB, small enough for smartphones, while retaining 98.2% of Qwen's benchmark scores (Julie Bort/TechCrunch)

Julie Bort / TechCrunch: PrismML releases Bonsai 2 27B, which compresses Alibaba's Qwen3.8 27B to 5.9 GB, small enough f…

◄ BACK TO ARTICLE