Artificial Analysis 发布端点准确率（Endpoint Accuracy）评测结果，覆盖 GLM-5.3 Flash、GLM-5.3、Muse Spark Open、DeepSeek V4.1 Flash 1017、Kimi K3 和 gpt-oss-120b 六款模型。各模型的完整结果可在其官网对应页面查看，评测方法详见官方 methodology 页面。
➤ Full results: GLM-5.3 Flash (https://artificialanalysis.ai/models/glm-5-3-flash/providers#endpoint-accuracy), Muse Spark Open (https://artificialanalysis.ai/models/muse-spark-open/providers#endpoint-accuracy), GLM-5.3 (https://artificialanalysis.ai/models/glm-5-3/providers#endpoint-accuracy), DeepSeek V4.1 Flash 1017 (https://artificialanalysis.ai/models/deepseek-v4-1-flash-1017/providers#endpoint-accuracy), Kimi K3 (https://artificialanalysis.ai/models/kimi-k3/providers#endpoint-accuracy) and gpt-oss-120b (https://artificialanalysis.ai/models/gpt-oss-120b/providers#endpoint-accuracy)
➤ Methodology: https://artificialanalysis.ai/methodology/endpoint-accuracy-index
来源：Artificial Analysis · x.com