Lightport, a lightweight AI gateway forked from Portkey, makes LLM providers OpenAI-compatible by transforming requests and responses.

Lightport accepts OpenAI-compatible requests, transforms them for the target provider, and returns the response. It supports 77 total providers including OpenAI, Anthropic, Azure OpenAI, Google Gemini, Vertex AI, Bedrock, Cohere, Mistral, Groq, Deepseek, Together AI, Fireworks, and Ollama.

The gateway implements request/response transformation only—retries, secret management, caching, rate limiting, and other operational concerns are explicitly non-goals and are better handled at a higher abstraction layer. It started as a fork of Portkey AI Gateway, since fixed numerous bugs, added integration tests for every provider, and continues to actively maintain the gateway as it directly powers Glama.

Supported endpoints include POST /v1/chat/completions, POST /v1/completions, and POST /v1/responses (+ GET, DELETE, input_items). The gateway runs on http://localhost:8787 and accepts provider specification via the x-lightport-provider header along with credentials passed via Authorization or provider-specific headers.