DeepSeek, an artificial intelligence startup based in Hangzhou, open-sourced DeepInfer on Oct. 20, SiliconANGLE reported.[1] DeepInfer is the inference framework behind DeepSeek's own production services, software for serving large models to the public.[1] DeepSeek has not released its key-value cache management and load-balancing code, which it says is tightly coupled to internal infrastructure.[1] The code is on GitHub under the Apache 2.0 license, though DeepSeek has not yet released its key-value cache management and load-balancing code.[1]