KTransformers

Popular Model Usage

Use this page as the model-family router. Exact validation status lives in the Support Matrix; individual model pages describe the current support scope and remaining validation needs.

Model familyCurrent entryNotes
DeepSeek V4-Flashkt run deepseek-v4-flash or manual MXFP4 launchNarrow path; verify package and attention constraints before production use.
DeepSeek V3.2kt run deepseek-v3.2 for registry FP8; tutorial path still mentions AMXINT4Needs reconciliation between registry default and tutorial method.
DeepSeek V3 / R1kt run deepseek-v3, kt run deepseek-r1Current registry exists; older DeepSeek pages using legacy servers are historical references.
Kimi K2 Thinkingkt run kimi-k2-thinking or RAWINT4 manual launchUse backend-specific behavior within the scope described on the model page.
MiniMax M2 / M2.1kt run m2, kt run m2.1Registry includes parser defaults and tensor-parallel constraints.
MiniMax M2.5Manual SGLang-KT tutorialNeeds minimum validation before use in the target environment.
Qwen3 / Qwen3.5 / Qwen3-Coder-NextManual BF16, FP8, or GPTQ_INT4 examplesChoose method by exact checkpoint and CPU backend.
GLM-5 / GLM-5.1Manual BF16, FP8, or FP8_PERCHANNEL examplesTreat transformer version constraints as part of the supported configuration.

Before Copying Commands

Commands are directly reusable when all of these match:

model family + checkpoint + KT method + CPU ISA/backend + GPU count + package versions

If any field changes, treat the command as a starting point and run a minimum validation on your target environment.