Qwen 3.8 27B scores 52 on the Artificial Analysis Intelligence Index
A 27B-parameter Qwen model is matching much larger frontier-class scores.
Willison notes Qwen 3.8 27B scored 52 on Artificial Analysis’s Intelligence Index, the same as GPT-5.6 Luna in max mode. It sits one point behind GLM-5.2 max and DeepSeek V4 Pro 0813 max, which he says are 753B and 1.7T parameters. His point is the size gap: Luna’s size is unknown, but likely far larger than 27B. Simon Willison's note
Willison notes Qwen 3.8 27B scored 52 on Artificial Analysis’s Intelligence Index, the same as GPT-5.6 Luna in max mode. It sits one point behind GLM-5.2 max and DeepSeek V4 Pro 0813 max, which he says are 753B and 1.7T parameters. His point is the size gap: Luna’s size is unknown, but likely far larger than 27B. Simon Willison's note
score 7