ParallelHTTP is a lightweight, zero-dependency tool for load testing and benchmarking HTTP APIs. Configure the number of concurrent requests, method, timeout, and duration — then get back structured results with latency percentiles and per-request status codes.
It ships as a single binary, a Docker image, and a Web UI — so you can use whichever fits your workflow.
| Parallel requests | Fire N requests simultaneously and collect every response |
| Latency percentiles | P50, P90, P99 aggregated across all requests |
| Multiple output formats | json, yaml, or text |
| Web UI | Visual interface for interactive testing in the browser |
| CLI | Scriptable, CI-friendly terminal usage |
| CSV export | Download results directly from the Web UI |
| Test endpoint | Built-in endpoint that returns random responses for trying things out |
The fastest way to get up and running:
# Run the Web UI with Docker — no installation needed
docker run --rm -p 8080:8080 nicumicle/parallelhttpThen open http://localhost:8080 in your browser.
Download the pre-built binary for your OS from the Releases page.
macOS / Linux:
# Make it executable
chmod +x parallelhttp
# Confirm it works
./parallelhttp --helpWindows:
Download parallelhttp.exe and run it from Command Prompt or PowerShell:
.\parallelhttp.exe --helpdocker run --rm -p 8080:8080 nicumicle/parallelhttpRequires Go 1.24+.
git clone https://github.com/nicumicle/parallelhttp.git
cd parallelhttp
# Build the binary
go build -o parallelhttp ./cmd/cli/main.goStart the server using any of the following methods:
# Using the binary
./parallelhttp --serve --port=8080
# Using Docker
docker run --rm -p 8080:8080 nicumicle/parallelhttp
# From source
go run cmd/service/main.goOpen http://localhost:8080 in your browser.
Results:
Run parallel requests directly from your terminal.
./parallelhttp \
--endpoint=https://api.example.com/health \
--parallel=10 \
--method=GET \
--timeout=2s \
--duration=30s \
--format=jsonFlags:
| Flag | Required | Description | Default |
|---|---|---|---|
--endpoint |
Yes | Target URL | — |
--parallel |
No | Number of concurrent requests | 1 |
--method |
No | HTTP method: GET POST PUT PATCH DELETE |
GET |
--timeout |
No | Per-request timeout (e.g. 500ms, 2s) |
10s |
--duration |
No | Total run duration (e.g. 30s, 5m). 0 = unlimited |
0 |
--format |
No | Output format: json yaml text |
json |
--serve |
No | Start the Web UI server | — |
--port |
No | Web UI server port | 8080 |
Example output (--format=json):
{
"requests": [
{
"response": {
"status_code": 200,
"time": "2025-12-02T04:39:26.450811Z",
"duration": 176680135,
"duration_h": "176.68ms"
},
"error": null,
"error_message": null
}
],
"stats": {
"start_time": "2025-12-02T04:39:26.450727Z",
"end_time": "2025-12-02T04:39:26.630824Z",
"duration": "180.09ms",
"latency": {
"p50": "177.10ms",
"p90": "179.94ms",
"p99": "179.94ms"
}
}
}When the server is running, a built-in test endpoint is available that returns random HTTP responses. It is useful for verifying your setup before pointing the tool at a real API.
curl http://localhost:8080/testImportant
ParallelHTTP is intended only for testing APIs, servers, and domains you own or have explicit permission to test.
Sending high-volume requests to systems without authorization may violate terms of service, computer misuse laws, or security policies. The authors assume no liability for any misuse.
Test only what you own.
Contributions are welcome. Please open an issue to discuss significant changes before submitting a pull request.
MIT — see LICENSE for details.