DeepSeek Open-Sources deepseek-harness & DeepSelect: DSA Sparse Attention and DeepGEMM for High-Throughput Inference

By Dr. Erik Vandeberg • Published on 2026-09-08 • 5 min read •Source: DeepSeek GitHub

DeepSeek open-sources its core production runtime 'deepseek-harness', combining Dynamic Sparse Attention (DSA) Top-K selection with custom FP8 DeepGEMM matrix kernels.

DeepSeek has published **deepseek-harness**, the inference and plugin orchestration harness powering its web and API infrastructure.

DeepSelect & DeepGEMM deepseek-harness open-sources DeepSelect—a GPU-accelerated routing kernel that evaluates attention sparsity in real-time, executing Dynamic Sparse Attention (DSA) with near-zero latency overhead. Paired with DeepGEMM FP8 matrix operations, self-hosted clusters achieve up to a 3.4× boost in tokens-per-second-per-GPU.

Source & Fact Check

This technical dispatch was verified against primary documentation released by DeepSeek GitHub.

Read Original Announcement on DeepSeek GitHub →