/**
 * kosha-discovery — DeepInfra provider discoverer.
 *
 * Queries the DeepInfra `/v1/models` endpoint (OpenAI-compatible),
 * filters relevant models, and maps them into {@link ModelCard} objects.
 *
 * DeepInfra's API catalog is multi-vendor, hosting models from meta, mistral,
 * deepseek, qwen, bigcode, google, and others. Model IDs use `vendor/model-name`
 * namespacing (e.g. `meta-llama/Meta-Llama-3.1-405B-Instruct`).
 * @module
 */
import { OpenAICompatibleDiscoverer, type OpenAICompatibleModel, type ModelClassification } from "./openai-compatible.js";
/**
 * Discovers models available through the DeepInfra API (api.deepinfra.com).
 *
 * The DeepInfra catalog serves models from multiple vendors using an
 * OpenAI-compatible API. Model IDs are namespaced (e.g.
 * `meta-llama/Meta-Llama-3.1-405B-Instruct`, `Qwen/Qwen2.5-72B-Instruct`),
 * allowing origin provider extraction from the prefix.
 */
export declare class DeepInfraDiscoverer extends OpenAICompatibleDiscoverer {
    readonly providerId = "deepinfra";
    readonly providerName = "DeepInfra";
    readonly baseUrl = "https://api.deepinfra.com";
    /**
     * Filter out models that are not suitable for direct inference.
     *
     * We exclude:
     * - Reward models (used for RLHF training pipelines, not inference)
     */
    protected isRelevantModel(model: OpenAICompatibleModel): boolean;
    /**
     * Classify a DeepInfra model: extract origin provider, infer mode and capabilities.
     *
     * DeepInfra uses `vendor/model-name` namespacing, so we extract the origin
     * provider from the prefix and apply well-known aliases to normalize names
     * (e.g. `meta-llama` -> `meta`, `mistralai` -> `mistral`).
     */
    protected classifyModel(model: OpenAICompatibleModel): ModelClassification;
}
//# sourceMappingURL=deepinfra.d.ts.map