Member Junction
    Preparing search index...
    • Picks the highest-power LLM that supports Image input (a vision-capable controller), or undefined when none qualify. Filters the candidate models to the LLM type, narrows to those for which supportsImageInput is true, then returns the one with the highest PowerRank (descending).

      Pure (no engine/singleton state) so it is unit-testable in isolation; MJComputerUseEngine's selectHighestPowerVisionLLM supplies AIEngine.Instance.Models + an AIEngine.ModelSupportsModality predicate. The original input array is not mutated.

      Parameters

      • models: MJAIModelEntityExtended[]

        The candidate models (typically AIEngine.Instance.Models).

      • supportsImageInput: (modelId: string) => boolean

        Predicate: does the model id accept Image input modality?

      Returns MJAIModelEntityExtended | undefined

      The best vision-capable LLM, or undefined when none qualify.