Qwen 3.8 Low and Medium Models Achieve Outstanding Benchmark Results

Independent AI evaluation platform Artificial Analysis has benchmarked the new Qwen 3.8 Low and Medium models, revealing exceptionally high performance scores. These results validate the capabilities of Alibaba's smaller-scale models in standard evaluations. The strong performance of these smaller models makes them highly attractive for local deployment, offering high efficiency without requiring massive computing resources. It also suggests that Qwen's previous successes were due to genuine architectural capability rather than over-optimization. The benchmarks were published by Artificial Analysis, which measures LLMs across quality, speed, and cost. The tested models belong to the Qwen 3 series, which includes both dense and Mixture of Experts (MoE) models.