home / testing / oha

oha

10k

HTTP load testing CLI for benchmarking APIs and web endpoints with concurrency, rate limits, and latency reports.

$brew install oha
Language
Rust
Stars
10,097
Category
Testing & QA
Agent
Ready
Agent Compatibility
JSON Output
Agent Skill
MCP Support
AI Analysis

oha is an HTTP load-testing CLI for generating concurrent traffic against a URL or a file of URLs and summarizing how the server responded. It covers quick benchmark runs, rate-limited load checks, and live terminal monitoring.

What It Enables
  • Benchmark APIs or sites with fixed request counts or timed runs, tuning concurrency, QPS, burst behavior, keep-alive, HTTP version, headers, auth, and request bodies.
  • Measure latency percentiles, histograms, first-byte timing, throughput, and status-code distribution, then export summaries as JSON or per-request data as CSV.
  • Replay more realistic traffic mixes with URL lists or generated URL patterns, and optionally persist successful request records to SQLite for later analysis.
Agent Fit
  • Non-interactive flags, --no-tui, and structured JSON or CSV output make it easy to slot into CI checks, performance regressions, and agent verify loops.
  • The default ratatui monitor is a real fullscreen TUI for humans watching a run, but the code uses a faster collection path when TUI is disabled, which is the better automation mode.
  • Fit is strongest for inspect-only HTTP performance checks; it does not model browser execution or richer multi-step user journeys.
Caveats
  • --output-format json produces a summary document rather than per-request event records; use CSV or the SQLite sink when each request matters.
  • It operates at the raw HTTP request layer, so JavaScript execution, rendering, and browser-managed session behavior are outside scope.