Portkey AI Gateway

Portkey's AI Gateway sits between your app and AI model providers. You send requests to one OpenAI-compatible API, and the gateway routes them to OpenAI, Anthropic, Bedrock, Groq and many other providers' language, vision, audio and image models. It's lightweight and starts locally with a single npx command.

The gateway prevents downtime with automatic retries and fallbacks between providers, scales apps with load balancing and conditional routing, and protects deployments with guardrails. It handles multi-modal requests, works with agent frameworks and libraries such as LlamaIndex, and includes an MCP Gateway for managing MCP servers with authentication and observability. A local console shows requests, and you can deploy it to your own infrastructure or use Portkey's hosted gateway.

Portkey's gateway gives a team one place to switch, balance and protect its model traffic without being locked into a single provider. It's MIT-licensed. Portkey is now part of Palo Alto Networks, and its commercial product has been renamed PRISMA AIRS AI Gateway.

Stars

0

Forks

0

Commits

0

Repository

Portkey-AI/gateway

• OpenRouterHosted API that routes requests to many AI models.
• Cloudflare AI GatewayCloudflare's hosted gateway for AI requests.
• Vercel AI GatewayVercel's hosted gateway for AI models.
• LiteLLMNo. 091Open-source proxy for calling many LLM APIs in one format.

See every open-source alternative to OpenRouter in the collection.