@squidcloud/client
    Preparing search index...

    Interface BulkIngestionProviderRouteThroughput

    Provider-side progress of one job's batches on one provider route (provider, endpoint and model). A route is the unit the provider queues, and one batch can carry several lanes, so progress is reported per route.

    interface BulkIngestionProviderRouteThroughput {
        batchesInFlight: number;
        completedPerMinute?: number;
        countsObservedAt?: Date;
        endpoint:
            | "/v1/embeddings"
            | "/v1/messages/batches"
            | "/v1/chat/completions"
            | "/v1/ocr";
        lanes: BulkIngestionRequestLane[];
        model: string;
        provider: "voyage"
        | "anthropic"
        | "openai"
        | "mistral";
        requestsCompleted: number;
        requestsFailed?: number;
        requestsInFlight: number;
    }
    Index
    batchesInFlight: number

    The route's batches created or submitted and not yet ended at the provider.

    completedPerMinute?: number

    Requests the provider completed per minute over the last BulkIngestionProviderThroughput.windowMinutes; absent when no batch of the route reports counts.

    countsObservedAt?: Date

    The oldest view of an in-flight batch: when its counts were read, else when its status last was, else when it was created. Older than a few minutes means polls lag or return no counts, not that the provider stalled: completedPerMinute then misses what the provider did since.

    endpoint:
        | "/v1/embeddings"
        | "/v1/messages/batches"
        | "/v1/chat/completions"
        | "/v1/ocr"

    The provider API the route's batches ride.

    The lanes the route's batches carry.

    model: string

    The model the route's batches run on.

    provider: "voyage" | "anthropic" | "openai" | "mistral"

    The provider running the route's batches.

    requestsCompleted: number

    Requests the provider ran to a result, succeeded or failed, across the route's batches.

    requestsFailed?: number

    Of requestsCompleted, the ones that failed; absent when no batch reports that breakdown.

    requestsInFlight: number

    Requests in those batches the provider has not settled yet.