This article reviews the performance of several small AI models priced under 10 yuan in literary polishing tasks. Kimi K2 Thinking wins with its exceptional results and cost-effectiveness, approaching the level of Gemini 3; while the Gemini 3 series has its merits, it suffers from serious hallucination issues that affect stability; other models like Grok, MiniMax, and GLM have problems such as output shortening, logical disruption, and copy-pasting. Based on actual testing, this review provides readers with a practical guide to choosing AI models, especially suitable for users interested in AI literary creation.
