ModelOps SDK

ModelOps is a TypeScript SDK that unifies OpenAI, Anthropic, and Gemini behind a single API while automatically tracking tokens, latency, estimated cost, and business context.

Build once. Switch providers. Understand your AI usage.

Why ModelOps

• One SDK for OpenAI, Anthropic, and Gemini
• Automatic token, latency, and cost tracking
• Business-aware analytics with request context
• Unified request and response API
• Zero custom observability pipeline

Note: Cost is estimated from public provider pricing and may differ from your invoice if you use enterprise pricing, discounts, cached tokens, batch APIs, or other provider-specific billing rules.

Why not use the provider SDKs?

You absolutely can. But once you need to support multiple providers or understand AI usage in production, you'll likely end up building much of the same infrastructure yourself.

Instead of building that yourself, ModelOps gives you:

• One API across multiple providers
• Built-in observability
• Token and cost tracking
• Business-level analytics
• A centralized dashboard

How it works

ModelOps wraps the official provider APIs.

Your application continues calling OpenAI, Anthropic, or Gemini through a single SDK, while ModelOps automatically collects telemetry and sends it to your dashboard.

No proxy or infrastructure changes required.

Installation

npm install model-ops

Environment Variables

Before using the SDK:

1. Create a ModelOps account at https://modelops.me
2. Copy your ModelOps API key from Account Settings
3. Set at least one provider API key

Provider keys can be loaded from environment variables:

OPENAI_API_KEY=your_openai_api_key
ANTHROPIC_API_KEY=your_anthropic_api_key
GEMINI_API_KEY=your_gemini_api_key

ModelOps also requires your platform API key:

MODELOPS_API_KEY=your_modelops_api_key

You can provide provider keys through environment variables or directly in modelKeys.

Quick Start

Send your first request:

import { ModelOps, ModelOpsChatRequest } from "model-ops";

const modelOps = new ModelOps({
	apiKey: process.env.MODELOPS_API_KEY!,
	modelKeys: {
		openai: process.env.OPENAI_API_KEY,
		anthropic: process.env.ANTHROPIC_API_KEY,
		gemini: process.env.GEMINI_API_KEY,
	},
	captureContent: false,
});

const request: ModelOpsChatRequest = {
	provider: "openai",
	model: "gpt-4.1-mini",
	messages: [
		{
			role: "user",
			content: "Tell me a joke",
		},
	],
};

const response = await modelOps.chat(request);
console.log(response.content);

Business Context (Recommended for production)

Business context helps you analyze usage by product surface, customer, and workflow instead of only tokens and requests.

const response = await modelOps.chat({
	provider: "openai",
	model: "gpt-4.1-mini",
	messages: [
		{ role: "user", content: "Plan a 3-day trip to Tokyo." },
	],
	context: {
		useCase: "trip-planning",
		feature: "generate-itinerary",
		customerId: "cust_123",
		sessionId: "sess_456",
		workflowId: "wf_789",
		environment: "production",
	},
});

What Is Tracked Automatically

Each request records:

• Provider
• Model
• Input tokens
• Output tokens
• Total tokens
• Estimated request cost
• Request latency
• Success or failure
• Optional business context

Privacy

By default, ModelOps sends only telemetry needed for observability.

Collected data includes:

• Provider
• Model
• Token usage
• Estimated cost
• Latency
• Request context (optional)

Prompt and response content is not collected unless content capture is enabled.

Your provider API keys are used only to send requests directly to the selected provider and are never stored by ModelOps.

const modelOps = new ModelOps({
	apiKey: process.env.MODELOPS_API_KEY!,
	captureContent: true,
});

Supported Providers

• OpenAI
• Anthropic
• Gemini

More providers are planned.

Dashboard Insights

ModelOps dashboard insights include:

• Cost per feature
• Cost per customer
• Cost per provider
• Cost per model
• Token usage trends
• Latency trends
• Error rates
• Request volume

Early Access Pricing

ModelOps is currently free while we're building the platform and gathering feedback from early adopters.

Paid plans will be introduced in the future for advanced capabilities such as longer data retention, team collaboration, optimization recommendations, alerts, and enterprise features.

We'll communicate any pricing changes well in advance.

Roadmap

Available today:

• Unified SDK
• OpenAI, Anthropic, and Gemini support
• Token tracking
• Estimated cost
• Latency tracking
• Business context
• AI usage dashboard

Coming soon:

• Prompt optimization recommendations
• Model recommendations
• Prompt caching suggestions
• RAG optimization
• Cost anomaly detection
• Automatic model routing

Links

• Product dashboard: https://modelops.me
• npm package: https://www.npmjs.com/package/model-ops
• GitHub repository: https://github.com/adipeleg/modelOps

License

ISC