Nodes/ComfyUI-Ollama-ImageList/Llama.cpp Native Speculative Config (Compat)
ComfyUI Node

Llama.cpp Native Speculative Config (Compat)

Bundles DFlash, DSpark, or Native MTP configuration and its optional draft GGUF into one typed connection.

By craftingmod·Created 19 days ago·Updated 9 days ago· 2
Llama.cpp Native Speculative Config (Compat)
    • speculative
    presetOff
    draft_model[no GGUF models found]
    custom_spec_typedraft-dflash
    custom_mtp_provideroff
    spec_n_max2
    spec_n_min0
    spec_p_min0.00
    CategoryOllama/llama_cpp/experimental

    Inputs (7)

    NameTypeDefaultDescription
    presetCOMBOOff7 options: Off, Muse Glimmer DFlash, Generic DFlash, Generic DSpark, Gemma 4 External MTP, Qwen 3.5 Internal MTP, +1
    draft_modelCOMBO[no GGUF models found]Required by DFlash, DSpark, and Gemma 4 external MTP; ignored by Off and Qwen 3.5 internal MTP.
    custom_spec_typeCOMBOdraft-dflash4 options: none, draft-dflash, draft-dspark, draft-mtp
    custom_mtp_providerCOMBOoff3 options: off, external_gemma4, internal_qwen35
    spec_n_maxINT21–64
    spec_n_minINT00–64
    spec_p_minFLOAT0.000–1

    Outputs (1)

    NameTypeDescription
    speculativeOLLAMA_IMAGE_LIST_LLAMA_CPP_SPECULATIVE_CONFIG