API Response Times Benchmark 2026
Enterno.io measured response times for 1000 popular public APIs (March 2026) from three geographies (Frankfurt, Virginia, Singapore). Median p50 = 180ms, p75 = 340ms, p95 = 880ms, p99 = 2.4s. Edge-computing APIs (Cloudflare Workers) dominate p99. REST slower than GraphQL in p50 (190 vs 140ms) due to typical over-fetching.
Below: key findings, platform breakdown, implications, methodology, FAQ.
Key Findings
| Metric | Pass / Value | Median | p75 |
|---|---|---|---|
| p50 (median) response time | 180ms | 180 | — |
| p75 response time | 340ms | — | 340 |
| p95 response time | 880ms | — | — |
| p99 response time | 2,400ms | — | — |
| APIs with p99 < 500ms | 18% | — | — |
| APIs with p99 > 5s | 9% | — | — |
| APIs using Cloudflare Workers (edge) | 22% | — | — |
| REST vs GraphQL p50 | 190 vs 140 ms | — | — |
Breakdown by Platform
| Platform | Share | Detail | — |
|---|---|---|---|
| Cloudflare Workers | 22% | p50: 45ms / p99: 210ms | — |
| AWS Lambda (EU-West) | 24% | p50: 110ms / p99: 1.2s | — |
| Google Cloud Run | 9% | p50: 140ms / p99: 1.8s | — |
| Vercel Edge Functions | 7% | p50: 60ms / p99: 340ms | — |
| Yandex Cloud Functions | 4% | p50: 180ms / p99: 2.1s | — |
| Classic VPS + PHP/Node | 28% | p50: 310ms / p99: 4.5s | — |
| Self-hosted (on-prem) | 6% | p50: 420ms / p99: 6.8s | — |
Why It Matters
- Edge computing (Cloudflare Workers) — the only path to < 100ms p50 at global scale
- For B2C SPAs targeting <100ms TTFB — Cloudflare Workers or Vercel Edge
- GraphQL 25% faster than REST at p50 thanks to under-fetching, but more complex to cache
- Cold start on Lambda can add up to 2s at low load — killer for real-time APIs
- p99 > 1s = user-noticeable slowdown. Target p99 < 500ms for mission-critical APIs
Methodology
Top-1000 public APIs from Postman Public API Network + awesome-lists. GET requests on standard endpoints with 100 replications, 5-minute intervals, 7 days. Measured via Enterno.io infrastructure (msk + frankfurt + virginia + singapore). REST vs GraphQL: where both available — tested both endpoints.
Learn more
Frequently Asked Questions
Why is p99 so important?
p50 — average user. p99 — worst 1% of requests. For SLAs (99% uptime + 99% fast) you need both. Typical p99 = 10-20× p50.
Cloudflare Workers vs Lambda — what's different?
Workers — V8 isolates (1-5ms cold start). Lambda — container (100ms-2s cold start). Workers win real-time, Lambda wins large compute.
How to measure response time without monitoring?
<code>curl -o /dev/null -s -w "%{time_total}\n" https://api.example.com</code> gives one sample. For statistics → <a href="/en/monitors">Enterno Monitor</a> every-minute checks.
What speeds up API responses?
1) Edge computing / CDN. 2) Async I/O (Node, Go, Rust). 3) Database indexes. 4) HTTP/2 or 3. 5) Payload compression.