GLM-5.3-Flash BIZ for DGX Spark-class GB10 systems. 3.x serves the XSFP4 checkpoint or NVIDIA's pinned NVFP4 one with TensorFold's native Zig/CUDA engine (TP=2 or TP=3, MTP drafts equal serial, tool calls, structured output, images, video). Apache-2.0 code, weights fetched separately. BIZ = business-use intent, not support or certification.
zig cuda nvidia arm64 quantization tensor-parallelism vllm llm-inference speculative-decoding gb10 nvfp4 dgx-spark glm-5-3-flash tensorfold xsfp4
-
Updated
Oct 23, 2026 - Python