In a UX Roundup for July 27, 2026, several high-end AI models were compared: OpenAI’s GPT 5.6 Sol, Anthropic’s Claude Opus 5, and Chinese Moonshot’s Kimi K3. When asked for suggestions on improving an article, Kimi K3 was the winner, suggesting a new concept that added true value, depth, and a new angle. GPT 5.6 Sol gave the most detailed but overly pedantic feedback, while Opus 5 was not truly insightful. For the author’s use cases, Kimi K3 is the best of these new models, but it still doesn’t reach the depth of insight from Claude Fable 5, which remains the leading frontier model. The author noted that different frontier models can be better for different purposes, as GPT 5.6 Sol is reportedly as good as Fable in agentic mode. Separately, Superdesign published 26 statistics drawn from 210,759 real prompts that users sent to its AI design agent from January to June 2026, based on logs of actual behavior. Additionally, the Delhi High Court ruled that AI training on copyrighted content is legal in India.