Recently, KAT-Coder-Pro V1, the Agentic Coding model developed by Kuaishou, announced a major upgrade and achieved outstanding results in the official evaluation of the renowned large model competition platform Artificial Analysis Intelligence Index.
KAT-Coder-Pro V1 scored 64 points, successfully surpassing Claude4.5Sonnet and strongly entering the top 10 of the overall ranking. More notably, the model secured the first place in the Non-Reasoning Model ranking with a significant advantage.
The evaluation results show that KAT-Coder-Pro V1 not only has excellent performance but also consumes far fewer output tokens than other models in the same performance range, achieving an excellent balance between price and performance.
