GenCost: Generative Content Economics Engine – Making AI API Costs Transparent and Controllable

GenCost is a real-time cost metering and optimization infrastructure designed specifically for generative AI applications. It supports core features like multi-model intelligent routing, cost attribution tracking, and waste detection, aiming to solve the black box problem of API call costs. This project is participating in the AI Tinkerers x ElevenLabs Generative Media Hackathon NYC 2026, competing in the 'Economics & Attribution Tracking' track and the 'Best Cost Cutter' special award category.

With the widespread use of APIs for large language models, speech synthesis, etc., developers often use multiple service providers like ElevenLabs and OpenRouter, each with different pricing models. Traditional cost management relies on post-hoc bill analysis, which cannot real-time perceive the cost of a single call nor easily track the content generation lineage. As a content economics infrastructure, GenCost is inserted between applications and AI APIs to achieve real-time cost metering, intelligent routing, and fine-grained attribution.

GenCost adopts a layered architecture with six modules working together:

1. Metering Proxy: Intercepts API requests and records estimated and actual costs, latency, etc.
2. Intelligent Router: Automatically selects the cheapest model based on quality thresholds, with built-in real-time pricing from multiple service providers.
3. Content Pipeline: Supports full metering of multi-modal generation processes.
4. Attribution Tracker: Creates cost fingerprints for content assets, recording generation lineage and remix history.
5. Waste Detector: Marks unused content and quantifies sunk costs.
6. Real-time Dashboard: Built with Next.js, uses WebSocket to push real-time cost events and intuitively display expenditure status.

The backend is based on Python 3.11+ and FastAPI framework, using SQLite as a zero-configuration database; the proxy layer uses httpx to implement asynchronous request interception and forwarding, supporting API access from multiple service providers. The frontend uses Next.js 15 + React 19 + Tailwind CSS v4, Recharts for data visualization, and WebSocket to ensure real-time cost push. The project provides a demo data seed script with 55 pre-configured pipeline records for quick demonstration.

GenCost solves multiple problems for AI teams:

• Cost Budget Control: Set limits and quality levels to automatically optimize model selection.
• Multi-model Strategy Optimization: Intuitively compare model cost-performance to find cost-effective options.
• Content Asset Valuation: Cost attribution data supports decisions such as creator revenue sharing and content pricing.
• Team Cost Awareness Cultivation: Real-time dashboard feedback enhances cost cognition.

GenCost is fully open-source (MIT license) with a clear code structure, reflecting a deep understanding of industry pain points. For AI developers, it allows quick setup of a demo environment (total budget of ~$15-20 to experience real API calls) or integration as a cost metering module, offering high practical value and reference significance.