[GLM-5.3-Flash Defeats DeepSeek-V4-Flash: Higher Scores, Nearly 3 Times Cheaper] Beating AI Newsflash: The newly open-sourced GLM-5.3 Flash by Zhipu directly takes on DeepSeek-V4-Flash. Both models focus on affordability, coding, and agent functionalities, but based on currently available comparative evaluations, Zhipu's model is stronger. On Terminal-Bench 2.1, GLM-5.3 Flash scored 84.3, while DeepSeek-V4-Flash scored 82.7; in DeepSWE, the scores were 63.4 versus 54.4, leading by 9 points. The former also natively supports image and video input. Pricing has been significantly reduced as well. GLM-5.3-Flash, during a limited-time 50% discount, costs $0.075 per million tokens for standard input and $0.25 for output. Even during DeepSeek V4 Flash's cheapest off-peak period, its prices are $0.22 for input and $0.66 for output. GLM's input is nearly 3 times cheaper, and its output is 2.6 times cheaper. After the two-week promotion ends, GLM's prices will revert to $0.15 for input and $0.5 for output, still lower than DeepSeek's off-peak rates. Currently, DeepSeek is only cheaper in terms of cache-hit pricing.