GLM Models
GLM support is currently inference-focused and depends on the exact environment configuration.
Inference
| Model | Entry | Method | Status |
|---|---|---|---|
| GLM-5 | Manual SGLang-KT tutorial | FP8 or BF16 | Needs smoke. |
| GLM-5.1 | Manual SGLang-KT tutorial | FP8, BF16, or FP8_PERCHANNEL depending on page | Needs isolated environment smoke. |
GLM-5.1 may require a specific Transformers stack. GLM-5 and GLM-5.1 should be read as separate validation combinations.
Source Material
doc/en/kt-kernel/GLM-5-Tutorial.mddoc/en/kt-kernel/GLM-5.1-Tutorial.md