Claude Penetrates Core Territory of Domestic Low-Priced AI Models, Triggering Sharp Intraday Drops in Zhipu AI and MiniMax Stocks

Insight Beating AI News Flash: Anthropic’s newly released Claude Haiku 5.5 is exerting intense competitive pressure—akin to a "kill line"—on Chinese domestic low-cost large language models (LLMs). On October 8, Hong Kong-listed LLM firms saw broad declines: MiniMax fell over 9%, Zhipu AI dropped more than 6%, while the Hang Seng Tech Index fell roughly 1% in the same period. Claude Haiku 5.5’s base API pricing is $0.1 per million input tokens and $0.5 per million output tokens. By comparison, DeepSeek V4.1 Flash charges $0.3 and $1.2 per million input/output tokens during peak hours, sliding to $0.15 and $0.6 during off-peak periods. Zhipu AI’s GLM-5.3 Flash has standard pricing of $0.15 and $0.5 per million input and output tokens respectively. Claude Haiku 5.5 already holds clear unit price competitiveness when input tokens do not exceed 100,000. Artificial Analysis awarded Claude Haiku 5.5 a composite intelligence score of 43, outperforming GLM-5.3 Flash (42), DeepSeek V4.1 Flash (39), and Xiaomi’s MiMo-V2.6-Flash (38). In the Terminal-Bench 4.0 programming benchmark, Claude Haiku 5.5 and GLM-5.3 Flash tied at 33%, ahead of DeepSeek (27%) and MiMo (23%). However, Claude Haiku 5.5’s strengths are primarily in its base pricing and performance under maximum inference load. It has higher token consumption, and long context inputs trigger a fivefold pricing surcharge.