OpenAI-Compatible API Proxy

A production-ready proxy that accepts standard OpenAI API requests, maps model names to upstream equivalents, and forwards them to a configurable backend — with full streaming support.

Base URL

https://tmpmanueluhn65--c80791de310a11f1a33142dde27851f2.web.val.run

Endpoints

Method	Path	Description
`GET`	`/api/v1/models`	List available models (OpenAI format)
`POST`	`/api/v1/chat/completions`	Chat completions (streaming & non-streaming)
`OPTIONS`	`*`	CORS preflight

Model Mapping

Alias	Upstream Model
`gpt-5`	`chat-gpt`
`gpt-5-turbo`	`chat-gpt`
`gpt-4o`	`chat-gpt`
`gpt-4-turbo`	`chat-gpt`
`claude-4.6-sonnet`	`anthropic-claude-sonnet-4-5`
`gemini-2.5-pro`	`google-gemini-2-5-pro`
(unknown)	`chat-gpt`

Usage

List models

curl https://tmpmanueluhn65--c80791de310a11f1a33142dde27851f2.web.val.run/api/v1/models

Chat completion (non-streaming)

curl -X POST https://tmpmanueluhn65--c80791de310a11f1a33142dde27851f2.web.val.run/api/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_UPSTREAM_TOKEN" \
  -d '{
    "model": "gpt-5",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

Chat completion (streaming)

curl -X POST https://tmpmanueluhn65--c80791de310a11f1a33142dde27851f2.web.val.run/api/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_UPSTREAM_TOKEN" \
  -d '{
    "model": "claude-4.6-sonnet",
    "messages": [{"role": "user", "content": "Hello!"}],
    "stream": true
  }'

Architecture

Rendering mermaid diagram...

Auth

Pass your upstream API token as a Bearer token. The proxy validates the header format and forwards the token to the upstream provider. No tokens are stored.

Streaming Strategy

When stream: true, the proxy:

Fetches the full response from upstream with stream: false
Splits the content into ~4-character chunks
Emits SSE events: first chunk (role), content chunks, finish chunk, [DONE]