How we built the fastest GLM 5 API

Introduced custom kernels for DeepSeek Sparse Attention (DSA) to optimize KV cache cost by using a lighting indexer to identify top-K tokens and perform sparse attention. Optimized the indexer code path by skipping it for sequences shorter than K and fusing operations for high-precision projections. Implemented a low-overhead MTP speculative decoding engine by running the draft and target models as one, minimizing host overhead. Optimized MoE dispatch kernels and quantized model weights to NVFP4 for Blackwell compatibility.