GLM-5.3-Flash-GGUF

GLM-5.3-Flash — Backpack GGUF

GGUF quantizations of zai-org/GLM-5.3-Flash, packaged for llama.cpp-compatible image-and-text inference and Backpack.

Model

Available packages

Memory values are estimates, not guarantees. Runtime configuration and context length change actual use.

Multimodal projector

The projector is required for image input and must be used with one of the language-model GGUF files above.

Backpack recommendation

Recommended: Q4_K_M. It usually offers a practical quality, size, and speed balance for local inference.

Run with llama.cpp

Using the llama.cpp revision recorded below:

llama-mtmd-cli --model GLM-5.3-Flash-Q4_K_M.gguf --mmproj GLM-5.3-Flash-mmproj-F16.gguf --image image.jpg --prompt "Describe this image."

Run with Backpack

These artifacts and backpack-model.yaml are prepared for the Backpack AI workspace.

Validation

Artifact integrity and GGUF metadata validation are the publication requirements. Runtime load, inference, and tokenizer results are reported independently and do not imply a certification or endorsement.

Runtime execution validation has not completed successfully for every artifact. Treat the affected package as experimental with the pinned toolchain until downstream runtime testing is complete.

• Packaged: 2026-09-06T07:07:13.528564+00:00

• llama.cpp toolchain revision: 8134115f88ed8018474e7db69afcfe97fb097fc4

• SHA-256 checksums: see checksums.sha256

• GLM-5.3-Flash-Q4_K_M.gguf: 3e1f1720e869d98acd55a8f94b5efd78814a6ba0a2c2e4e609d637e9cca60406

• GLM-5.3-Flash-mmproj-F16.gguf: f64a2e935c899224054258d2372d9ad4c19b760141044292fa6ee0fb5ff36624

Provenance

The source model was resolved to immutable revision 690b705278a3a58e538fcb37c2ca8b5f9511213c. It was converted with llama.cpp's convert_hf_to_gguf.py, including its multimodal projector, and quantized with llama-quantize; the exact toolchain revision is recorded above and in backpack-model.yaml.

License and attribution

Upstream declares mit. Review the upstream model card and comply with all applicable terms.

Backpack does not claim ownership of the original model. These artifacts are packaged and quantized distributions of the upstream model.

Disclaimer

Quantization can alter output quality. Memory estimates vary with runtime configuration, context length, and hardware.