Why does DeepSeek V4 Flash have such a low request limit for agentic workloads?
DeepSeek V4 Flash seems designed for agentic applications, where one task can easily trigger anywhere from 1 to 10 model calls. It is fast and inexpensive, but deployments…