kapynAI / Models

Qwen 3.8 27B scores 52 on the Artificial Analysis Intelligence Index

Qwen 3.8 27B matches GPT-5.6 Luna on the Artificial Analysis Intelligence Index. The 27-billion-parameter open-weights model achieves a score of 52, tying with much larger models like GPT-5.6 Luna and trailing massive trillion-parameter architectures by only a single point. This milestone highlights the rapid efficiency gains in mid-sized models, offering developers near-frontier performance at a fraction of the compute footprint.

Simon Willison·Aug 17, 2026

Opening Kapyn…