apinode-pro/ai-api-gateway-benchmark

A small, reproducible benchmark for OpenAI-compatible AI API gateways. It’s intended for developers and operators who need a repeatable health/performance check of an AI endpoint (default target: API NODE at https://apinode.pro, model gpt-5.5, OpenAI Responses API). Key features: - Measures success rate, average latency, p50 and p95, HTTP status, and whether output text is present. - Outputs: SUMMARY.md, TREND.md, results/last-run.(json|md) and dated history files. - CI integration: GitHub Actions runs scheduled/manual workflows and commits results back to the repo (requires APINODE_API_KEY secret). - Local run: simple npm script with env vars (base URL, key, model, sample count). - Guidance and docs: interpreting results, incident runbook, watch procedures, and reproducibility notes (include timestamp, region, sample size, timeout, model, endpoint, streaming vs non-streaming). Failure guidance lists common causes (expired key, rate limits, model mismatch, response shape changes) and advises checking last-run and CI logs.

AI-related keywords found in this repo's description, topics, or README summary — grouped by category. Each badge links to the corresponding ranking detail page.