---
type: "Article"
title: "GLM-5.3-Flash 320B UD-Q2_K_XL with llama.cpp on 1× DGX Spark"
description: "GLM-5.3-Flash 320B UD-Q2_K_XL with llama.cpp on 1× DGX Spark: recipe by Weschera with speed, condition and source for every number, weights and license."
resource: "https://www.contextstudios.ai/local-ai/weschera--glm53-flash-ud-q2kxl-mtp-1x-dgx-spark"
language: "en"
generated:
  by: "process:contextstudios-md/1"
  at: "2026-10-02T21:59:44.323Z"
status: "stable"
---

# GLM-5.3-Flash 320B UD-Q2_K_XL with llama.cpp on 1× DGX Spark

GLM-5.3-Flash 320B UD-Q2_K_XL with llama.cpp on 1× DGX Spark: 17.9 tok/s according to github.com (dataset as of Sep 29, 2026).

- Creator: [Weschera](https://www.contextstudios.ai/local-ai/creators/weschera.md)
- Engine: llama.cpp
- Quantization: Unsloth UD-Q2_K_XL (109 GB, ~2,4-bit Experts dynamisch; Attention/Router höher)
- Model family: GLM-5.3-Flash
- Hardware: dgx-spark x1
- Context: 262144

## All numbers

- 17.9 tok/s (Everyday; Everyday, 1 request, prose prompt, context 32k, no speculative decoding) — [source](https://github.com/Weschera/glm53-flash-one-spark/blob/main/README.md)
- decode_tps: 17.9 (1 request) — [source](https://github.com/Weschera/glm53-flash-one-spark/blob/main/README.md)
- decode_tps: 33.8 (1 request) — [source](https://github.com/Weschera/glm53-flash-one-spark/blob/main/README.md)
- prefill_tps: 291 (1 request) — [source](https://github.com/Weschera/glm53-flash-one-spark/blob/main/README.md)

## Weights

- [unsloth/GLM-5.3-Flash-GGUF](https://huggingface.co/unsloth/GLM-5.3-Flash-GGUF) (main, mit)
- [zai-org/GLM-5.3-Flash](https://huggingface.co/zai-org/GLM-5.3-Flash) (base, mit)

[https://github.com/Weschera/glm53-flash-one-spark](https://github.com/Weschera/glm53-flash-one-spark)

## Related

- [Local AI on your hardware.](https://www.contextstudios.ai/local-ai.md)