One key · 5 modalities · 30 models
Chat, image, video, speech, and transcription — one OpenAI-compatible API, one bill, and the real cost in every response.
# streaming chat completion
$ curl .../v1/chat/completions -d '{"model":"claude-4-7-sonnet"...}'
data: {"choices":[{"delta":{"content":"你好"}}]}
data: {"choices":[{"delta":{"content":" 👋"}}]}
data: [DONE]
x-relayx-model: anthropic/claude-4-7-sonnet
x-relayx-cost-usd: 0.000002 ← what this call costReal output
Real output from real calls — image, video, and speech, all made through RelayX.

Real-time synthesis, straight from /v1/audio/speech.
Five routes
curl https://relayx.timor419.com/v1/chat/completions \
-H "Authorization: Bearer $RELAYX_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "anthropic/claude-4-7-sonnet",
"messages": [{"role":"user","content":"Hi"}],
"stream": true
}'Read X-RelayX-Cost-Usd on every call; reconcile any window with GET /v1/usage.
Swap base_url and your existing SDKs keep working.
Frontier models stay reachable from mainland China.
Images, audio and video come back as URLs on our own R2 — your code never sees an upstream link.
Pay per call
Five minutes
Sign in with Google. Top up. Send your first request.