DeepSeek open-sources its DeepInfer inference engine

DeepSeek has open-sourced DeepInfer, the high-performance inference framework behind its large-scale production services, with this release focused on offline inference. It supports DeepSeek-V4.1-Flash, offers parallelism such as attention DP and MoE EP with EPLB, optimized kernels including FlashMLA and DeepGEMM, and DSpark draft-token decoding with confidence-scheduled verification.

• Western Labs Shift to Adopting Chinese AI Advances2026-09-30
• DeepSeek Raises $15 Billion to Challenge Leading US Labs2026-10-22
• China protests US DeepSeek ban while blocking rivals at home2026-10-21
• Vals AI: DeepSeek V4.1 Flash beats GPT 6 Sol on cybersecurity task2026-10-20
• Open GPU kernel engineering series covers GEMM to FlashAttention2026-10-19