Competitive benchmarks¶
The reproducible 2026-07-28 same-workload baseline used Hayate 0.15.1, FastAPI 0.140.0, Django 6.0.7, and Hono 4.12.32.
| Boundary | Hayate | FastAPI | Django | Hono |
|---|---|---|---|---|
| Cold start | 149.0 ms | 471.4 ms | 392.6 ms | 61.3 ms |
| Production packages | 5 | 13 | 6 | 2 |
| gzip payload | 298.1 KiB | 2,802.2 KiB | 5,147.2 KiB | 281.5 KiB |
| Throughput | 14,906 req/s | 10,086 req/s | 2,557 req/s | 59,187 req/s |
| Shared HTTP contract | 14/14 | 12/14 | 12/14 | 12/14 |
On this workload Hayate delivered 1.48× FastAPI's and 5.83× Django's throughput. Hono delivered 3.97× Hayate's throughput and retained the best startup, dependency-count, and deployment-payload results.
What this does not prove¶
- It does not rank every endpoint shape or deployment platform.
- The 14 checks are not a universal standards-compliance percentage.
- The local machine result is not a Cloudflare or AWS production benchmark.
- Ecosystem size and production history are adoption factors outside the request-per-second measurement.
Read the immutable publication and raw methodology
The separate capability matrix records feature evidence without inventing a weighted universal winner.