Z.ai announced GLM-5.3-Flash on August 26, 2026 as the named product behind the Ox Alpha stealth preview on OpenCode and OpenRouter. It is a 320 billion parameter mixture-of-experts-style model with 18 billion active parameters, 1M-token multimodal context, and weights released under MIT license at huggingface.co/zai-org/GLM-5.3-Flash. Z.ai claims the entire stealth week ran on Chinese AI chips, and published agent benchmarks competitive with Claude Opus 4.8 on GDPVal-AA v2 and DeepSWE. Standard API pricing is $0.15/M input, $0.50/M output, and $0.03/M cached input. It is a distinct SKU from GLM-5.3 (743B-base cyber-defense line launched August 14, 2026).