GLM OCR OQ8 FP16
This model was quantized using oQ mixed-precision quantization.
Model source
- Kind
- Text model
- File size
- 1.48 GB
- Base model
- zai-org/GLM-OCR
- Version
- vmodel
- Updated
- Nov 2, 2026
- Downloads
- 30
- Source
- Hugging Face
Source description
GLM-OCR-oQ8-fp16
This model was quantized using oQ mixed-precision quantization.
float16 gives ~20% faster prefill on M1/M2 Apple Silicon (native fp16). bfloat16 is safer on M3/M4 and for numerical stability.
Benchmark (on M1 Max)
| Model Variant | PP (Tokens per second) | TG (Tokens per second) |
|---|---|---|
| Original (bf16) | 4,684 | 104.8 |
| oQ8-fp16 | 3,806 | 99.0 |
Sources
1 source
Sources
1 sourceVerified Nov 2
Download files
Repository files
File inventory only; completeness is unverified.
Model artifacts
1 artifact
Model artifacts
1 artifactSource excerpts
3 excerpts
Source excerpts
3 excerpts--- license: mit base_model: - zai-org/GLM-OCR pipeline_tag: image-to-text language: - zh - en - fr - es - ru - de - ja - ko tags: - mlx - oq - quantized library_name: mlx --- # GLM-OCR-oQ8-fp16 This model was quantized using [oQ](https://github.com/jundot/omlx) mixed-precision quantization. float16 gives ~20% faster prefill on M1/M2 Apple Silicon (native fp16). bfloat16 is safer on M3/M4 and for numerical stability. # Benchmark (on M1 Max) | Model Variant | PP (Tokens per second) | TG (Tokens per second) | | :--- | :---: | :---: | | Original (bf16) | 4,684 | 104.8 | | oQ8-fp16 | 3,806 | 99.0 |
Source context: 30 downloads · 3 likes · Pipeline image-to-text · Library mlx · Repo RepublicOfKorokke/GLM-OCR-oQ8-fp16
Source context: 69 downloads · 3 likes · Pipeline image-to-text · Library mlx · Repo RepublicOfKorokke/GLM-OCR-oQ8-fp16
