Learn about this first
DeepSeek announces that it will fully open-source its internal reasoning framework DeepInfer, while also open-sourcing the production-grade Token encoding and decoding component deepseek-recipe.
Latest developments
SGLang vs. DeepInfer: measured DeepSeek-V4.1-Flash inference performance and full serving capabilities
Event progress3
SGLang integrates DeepInfer for optimization and compares performance
LMSYS Blog: SGLang Compared with DeepInfer: Real-World Inference Performance and Full Serving Capabilities of DeepSeek-V4.1-FlashThe vLLM team shares research on optimizing the large-scale serving performance of DeepSeek-V4.1-Flash
Official vLLM Blog: Serving DeepSeek-V4.1-Flash at Scale on GB300 NVL72, with Decoding Throughput Within 6% of DeepInferDeepSeek open-sources the inference framework DeepInfer
Official WeChat account: DeepSeek: DeepSeek Fully Open-Sources Its Internal Inference Framework DeepInfer