Lightport – AI gateway that makes LLM providers OpenAI-compatible
Stripped-down Portkey fork handling protocol translation for 77 providers without enterprise bloat.

Open-source proxy making 77+ LLM providers OpenAI-compatible without the enterprise bloat.
Backend developers building multi-provider AI applications
LiteLLM · Portkey
I am the founder of Glama.
We are making Lightport open-source – it's the AI gateway that's been powering Glama.
GitHub: https://github.com/glama-ai/lightport
Live: https://glama.ai/ai/gateway
Why?
We're going all-in on the MCP ecosystem – it's what we're best at. Open-sourcing the gateway is both a thank-you to the community that helped us grow and a way to keep us focused.
The short backstory:
Lightport began as a fork of Portkey. We needed a way to make various LLM providers OpenAI-compatible, and Portkey provided a solid foundation. But it also came with many higher-level features (guardrails, billing, etc.) that we didn't think belonged at this layer – and that made it hard to iterate on provider compatibility. So we slimmed it down, fixed bugs, added integration tests for 80+ providers, and shaped it into one thing: a reliable, lightweight layer that makes any LLM provider OpenAI-compatible.
What's next:
More modules will follow – guardrails, billing, retries, telemetry – each open-sourced as standalone middleware. We'll continue to maintain Lightport in the open, with a focus on OpenAI compatibility across LLM providers.
For Glama users:
The Glama AI gateway will continue to function as a privacy-first gateway, but we will support only a curated set of providers (OpenAI, Anthropic, Google, Grok, Groq, DeepSeek, Alibaba, Moonshot, and a few others). For everything else, OpenRouter is a great alternative.
Try it, break it, build with it. I can't wait to see what you make.
Stripped-down Portkey fork handling protocol translation for 77 providers without enterprise bloat.
LiteLLM already does this with more providers, more features, and way more maturity.
Go gateway with circuit breakers, but auth isn't production-ready yet.
Secret redaction and mlock security for AI gateways when LiteLLM already exists.
Yet another OpenAI-compatible gateway when LiteLLM and OpenRouter already exist.
Runs as a single binary with embedded SQLite and zero-config start, acting as a transparent, provider-agnostic proxy that logs model, tokens, latency, cost and API key hashes while leaving full body capture opt-in. It also proxies streaming responses in real time and exposes stable JSON analytics endpoints — a practical, instrumentable way to get reproducible, audit-ready traces for real LLM traffic, though long-term value depends on how it handles provider edge-cases and SDK compatibility.