/**
 * Bench - Main benchmarking class for PraisonAI Bench TypeScript
 *
 * This module provides the core benchmarking functionality using agents
 * to evaluate LLM performance across different tasks and models.
 */
import { TestResult } from './agent';
import { ProviderName } from './providers';
import { CostSummary } from './cost-tracker';
export interface BenchConfig {
    default_model?: string;
    default_provider?: ProviderName;
    output_format?: 'json' | 'csv';
    save_results?: boolean;
    output_dir?: string;
    max_retries?: number;
    timeout?: number;
}
export interface TestConfig {
    name?: string;
    prompt: string;
    model?: string;
    language?: string;
    expected?: string;
}
export interface TestSuiteConfig {
    config?: Record<string, unknown>;
    tests: TestConfig[];
}
export interface BenchSummary {
    total_tests: number;
    successful_tests: number;
    failed_tests: number;
    success_rate: string;
    models_tested: string[];
    average_execution_time: string;
    timestamp: string;
    cost_summary?: CostSummary;
}
export interface BenchResult extends TestResult {
    evaluation?: Record<string, unknown>;
    language?: string;
}
/**
 * Main benchmarking class that orchestrates agents for comprehensive LLM testing.
 */
export declare class Bench {
    private results;
    private config;
    private costTracker;
    private pluginManager;
    constructor(config?: BenchConfig, enableEvaluation?: boolean);
    /**
     * Detect language from response or test configuration.
     */
    private detectLanguage;
    /**
     * Run a single benchmark test.
     */
    runSingleTest(prompt: string, model?: string, testName?: string, expected?: string, language?: string): Promise<BenchResult>;
    /**
     * Run a complete test suite from a YAML or JSON file.
     */
    runTestSuite(testFile: string, testFilter?: string, defaultModel?: string, _concurrent?: number): Promise<BenchResult[]>;
    /**
     * Run the same test across multiple models for comparison.
     */
    runCrossModelTest(prompt: string, models?: string[]): Promise<BenchResult[]>;
    /**
     * Save benchmark results to file.
     */
    saveResults(filename?: string, format?: 'json' | 'csv'): string | null;
    private saveResultsCsv;
    /**
     * Generate HTML report from results.
     */
    generateReport(outputPath?: string): string | null;
    /**
     * Get a summary of benchmark results including costs.
     */
    getSummary(): BenchSummary;
    /**
     * Get all results.
     */
    getResults(): BenchResult[];
    /**
     * Clear all results.
     */
    clearResults(): void;
}
//# sourceMappingURL=bench.d.ts.map