OpenAI-compatible relay checker
AI API Relay Health Check
Enter a Base URL and API key. Fetch models, choose one, then run a focused quality test.
-
quality score
Result
Not tested
Fetch models, choose one, then start the test.
Question bank and per-question results
Latency
-
TTFT
-
Streaming total
-
Tokens/sec
-
Input tokens
-
Output tokens
-
Notes
Advanced raw responses
Request payload
{}
/v1/models
{}
/v1/chat/completions
{}
/v1/chat/completions (stream)
{}
What it checks
Model listing, selected model availability, control-question answer, OpenAI-style response shape, streaming quality, latency, TTFT, and rough token speed.
What it cannot prove
This is a practical quality signal, not an absolute audit. Slow, broken, or inconsistent behavior is a warning before you buy or rely on a relay.