Artificial Analysis announced that the Intelligence Index score for Qwen3.8 27B (xhigh) is 52. This result significantly exceeds the median score of 9 for models in the same class.
However, the model is considered expensive in terms of cost. Pricing is $0.50 per million input tokens and $3.00 per million output tokens. With the median prices being $0.05 and $0.17 respectively, it was reported that the evaluation cost $693.67 to conduct.
In terms of speed, it delivers 47 tokens per second, placing it in the slower category for its class. Additionally, the number of tokens generated during evaluation reached 160 million, showing a highly redundant output tendency compared to the median of 48 million.
This model supports text, image, and video inputs, with a context window of 256k tokens. It should be noted that this is the version with reasoning capabilities, and a non-reasoning version may also exist.
Source: Qwen3.8 27B scores 52 on Artificial Analysis（HN 381pt・180コメント） (HN Search (backfill), 2026-08-18)