What is Portkey?

Portkey is a comprehensive production stack designed for AI teams to build, deploy, and manage generative AI applications at scale. It integrates an AI Gateway, observability, guardrails, governance, and prompt management into a single platform. The AI Gateway provides a unified API to access over 1,600 LLMs, while observability offers real-time monitoring and debugging tools. Guardrails ensure AI outputs are safe and compliant, and prompt management streamlines prompt creation and versioning.

How to use Portkey?

To use Portkey, developers integrate it into their existing stack by adding just a few lines of code. They can start by signing up for a free account, then use the Portkey SDK (e.g., in Node.js or Python) to route requests through the AI Gateway. The platform provides a dashboard for monitoring, configuring guardrails, and managing prompts. Teams can also set up virtual keys for secure LLM access and use the observability module to track performance and costs.

Core features of Portkey

AI Gateway: Unified API to access 1,600+ LLMs with fallbacks, load balancing, retries, and caching.
Observability: Real-time logging, tracing, and analytics with 40+ metrics for cost, performance, and accuracy.
Guardrails: PII redaction, regex matching, and integration with providers like Patronus AI and AWS Bedrock.
Prompt Management: Versioning, playground, and API endpoints for prompt templates.
Governance: Role-based access control, SSO, and key management for enterprise security.
Model Catalog: Discover and manage models across providers.
MCP Gateway: Centralized authentication and observability for MCP servers.