Skip to content

LlamaParams

Per-model llama-server load parameters (the Model Hub “Load settings” panel). Absent fields fall back to the server’s own defaults.

optional batchSize?: number


optional cacheTypeK?: string


optional cacheTypeV?: string


optional ctxSize?: number


optional enableThinking?: boolean

Reasoning models only: false disables thinking (faster, less memory).


optional flashAttn?: boolean


optional mlock?: boolean


optional mmap?: boolean


optional nGpuLayers?: number


optional parallel?: number


optional seed?: number


optional threads?: number