I do not have verified information on specific coding performance benchmarks or differences between "GLM 5.2" and "Kimi K3." While I can confirm that Riven routes access to both GLM and Kimi models alongside others like Claude and GPT, I cannot provide verified technical comparisons, latency data, or coding proficiency rankings for these specific versions or models. For precise model capabilities, I recommend checking the official documentation from their respective providers or testing them directly via the Riven model picker on the homepage.