Benchmarks

Last updated: 2026-06-08

All figures come from one lab setup (CPU, OS, Node, tool versions). Your hardware will give different absolute RPS, but relative rankings stay consistent. Use the same scripts for your own numbers. In production, databases and external calls usually dominate latency — not the framework choice.

Overview

The benchmark suite in apps/benchmark supports 2 tools that produce slightly different absolute numbers but the same relative rankings:

The runner auto-detects wrk first; use --tool wrk|autocannon to force a specific tool.

Tool Comparison: NextRush v3

How the same test scenarios differ between wrk and autocannon on identical hardware (64 connections, 10s):

The numbers are within ~4% across all scenarios, so relative framework rankings stay consistent regardless of which tool you use.

Framework Comparison (autocannon)

Head-to-head comparison against popular Node.js frameworks. All servers implement identical endpoints with equivalent work.

Test Environment

Detailed Scenarios

app.get('/', (ctx) => ctx.json({ message: 'Hello World' }));

app.get('/users/:id', (ctx) => ctx.json({ id: ctx.params.id }));

app.post('/users', (ctx) => ctx.json(ctx.body, 201));

Deep Profile (wrk)

NextRush v3 benchmarked with wrk — a C-based HTTP benchmark tool that runs as a separate process, isolated from Node.js. Unlike autocannon, wrk doesn't share the event loop with the server, which eliminates the "testing yourself with yourself" problem.

Methodology

Scenario Breakdown

NextRush scales efficiently — the middleware stack pre-compilation means it can outperform even the hello-world baseline under concurrency.

Memory Profile

The numbers above include the full benchmark harness (wrk + server + monitoring). NextRush's own idle footprint is under 200KB.

Design Trade-offs: Fastify vs NextRush

Fastify edges ahead on raw throughput in these benchmarks. Here is what drives the gap:

Run Your Own Benchmarks

Quick comparison

cd apps/benchmark
pnpm install
pnpm bench:compare:quick

Defaults to wrk if installed, otherwise falls back to autocannon.

Force a specific tool

# Use autocannon
node scripts/run.js --compare --tool autocannon

# Use wrk with full profile (60s, 4 concurrency levels, 5 runs)
node scripts/run.js --profile full --tool wrk

Results are saved to results/ with timestamps. Each run generates a JSON file and a human-readable report.

How to use these numbers

Fastify edged ahead on mean RPS in this suite. NextRush came second. That ordering is useful for relative comparison on identical hardware — not as a universal ranking.

1. Fastify tunes aggressively for Node.js throughput. Its serialization choices pay off in micro-benchmarks.
2. In real apps, a single database call adds milliseconds per request. That gap between frameworks shrinks fast.
3. NextRush keeps a dependency-free core, ships adapters for several runtimes, and supports optional DI and decorators. Different trade-offs, not a single winner.

Run your own benchmarks on your own hardware. That is the only number that matters.