Huatai Securities Says Domestic AI Model Gap Has Narrowed To 3-4 Months

Huatai Securities said July updates to Kimi and the new GLM model released in the second quarter again confirmed the core logic behind domestic models, with iteration cycles likely to shorten further. According to 36Kr, the brokerage said the gap with overseas closed-source models has narrowed to 3-4 months, while domestic developers are improving capability ceilings and pricing power on the back of strong cost efficiency. Its testing found K3 only slightly weaker than GPT-5.6 Sol, and GLM 5.2 was positioned against Opus 4.7 after release. Huatai Securities said faster capability gains in domestic models could reinforce positive industry-chain trends, though competition at the model layer has intensified in the short term. In the long run, rising agent penetration could further expand the market opportunity.