New Features

SDK	Feature Description	Product Model
Vision SDK	None	Atlas 300I inference card (model 3010) (x86_64) Atlas 300I inference card (model 3000) (Arm) Atlas 500 A2 edge station Atlas 300I Pro inference card Atlas 300I Duo inference card Atlas 300V video analysis card Atlas 300V Pro video analysis card Atlas 200I SoC A1 core board Atlas 800I A2 inference product
Index SDK	Enhanced standard-state ILFlat algorithm IVFFlat retrieval algorithm supported IVFSP algorithm's Add interface performance optimized for small-batch scenarios	Atlas 300I Pro inference card Atlas 300V video analysis card Atlas 300V Pro video analysis card Atlas 300I Duo inference card Atlas 200I SoC A1 core board Atlas 300I inference card (model 3000) Atlas 300I inference card (model 3010) Atlas 800I A2 inference server
RAG SDK	Smart markdown document parsing supported Reference design for knowledge QA applications provided Embedding/Reranker model accelerated	Atlas 300I Duo inference card Atlas 800I A2 inference server
Rec SDK	Gradient accumulation added to the Torch scenario, supporting PyTorch 2.7 and DP parallel training; pagedHSTU supported in the inference scenario, with head_num configurable up to 16, different QKV lengths, and custom mask.	Atlas 800T A2 training server Atlas 200T A2 Box16 heterogeneous subrack Atlas 800T A3 SuperPoD server

Parent topic: Update Description