vitest-dev/vitest · error · TypeError

${matcherHint('.toBeSlowerThan')} expects the expected value

Error message

${matcherHint('.toBeSlowerThan')} expects the expected value to be a benchmark result.

What it means

Thrown by the `.toBeSlowerThan` benchmark matcher when the EXPECTED (second) argument is not a valid `BenchResult`. A valid result is a non-null object with a numeric `latency.mean` (produced by `bench()`/`vitest bench`). The matcher cannot compute a threshold without comparable latency data, so it refuses to proceed rather than returning a misleading pass/fail.

Source

Thrown at packages/vitest/src/integrations/chai/bench.ts:61

          + `Expected: ${EXPECTED_COLOR(formatOps(expected.throughput.mean))} ops/sec\n`
          : `${matcherHint('.toBeFasterThan')}\n\nExpected to be faster${delta > 0 ? ` by at least ${(delta * 100).toFixed(0)}%` : ''}, but was ${Number(relation) > 0 ? `${relation}% slower` : `only ${Math.abs(Number(relation))}% faster`}.\n\n`
            + `Received: ${RECEIVED_COLOR(formatOps(actual.throughput.mean))} ops/sec\n`
            + `Expected: ${EXPECTED_COLOR(formatOps(expected.throughput.mean))} ops/sec\n`
      },
    }
  },

  toBeSlowerThan(actual: unknown, expected: unknown, options?: { delta?: number }) {
    const { matcherHint, RECEIVED_COLOR, EXPECTED_COLOR } = this.utils
    const delta = options?.delta ?? 0

    if (!isBenchResult(actual)) {
      throw new TypeError(
        `${matcherHint('.toBeSlowerThan')} expects the actual value to be a benchmark result.`,
      )
    }
    if (!isBenchResult(expected)) {
      throw new TypeError(
        `${matcherHint('.toBeSlowerThan')} expects the expected value to be a benchmark result.`,
      )
    }

    const threshold = expected.latency.mean * (1 + delta)
    const pass = actual.latency.mean > threshold

    return {
      pass,
      message: () => {
        const relation = ((actual.latency.mean - expected.latency.mean) / expected.latency.mean * 100).toFixed(2)
        return pass
          ? `${matcherHint('.not.toBeSlowerThan')}\n\nExpected to not be slower, but was ${relation}% slower.\n\n`
          + `Received: ${RECEIVED_COLOR(formatOps(actual.throughput.mean))} ops/sec\n`
          + `Expected: ${EXPECTED_COLOR(formatOps(expected.throughput.mean))} ops/sec\n`
          : `${matcherHint('.toBeSlowerThan')}\n\nExpected to be slower${delta > 0 ? ` by at least ${(delta * 100).toFixed(0)}%` : ''}, but was ${Number(relation) < 0 ? `${Math.abs(Number(relation))}% faster` : `only ${relation}% slower`}.\n\n`
            + `Received: ${RECEIVED_COLOR(formatOps(actual.throughput.mean))} ops/sec\n`
            + `Expected: ${EXPECTED_COLOR(formatOps(expected.throughput.mean))} ops/sec\n`

View on GitHub (pinned to d568f8ce37)

Solutions

  1. Pass a full `BenchResult` object as the expected value, produced by another `bench()`/`vitest bench` task.
  2. If you want a numeric latency comparison, write a custom assertion or use `expect(actual.latency.mean).toBeGreaterThan(expectedMs)` instead.
  3. Verify the expected value has `typeof expected.latency.mean === 'number'` before calling the matcher.

Example fix

// before
expect(result).toBeSlowerThan(100)
// after
const baseline = await bench('baseline', baselineFn)
expect(result).toBeSlowerThan(baseline, { delta: 0.1 })
Defensive patterns

Strategy: type-guard

Validate before calling

function isBenchResult(v: unknown): v is { latency: { mean: number }; throughput: { mean: number } } {
  return typeof v === 'object' && v !== null
    && typeof (v as any).latency?.mean === 'number'
}

Type guard

import { bench } from 'vitest'
// vitest does not export isBenchResult publicly; use the local shape check above
const ok = isBenchResult(expected)
if (!ok) throw new Error('expected is not a bench result')
expect(actual).toBeSlowerThan(expected)

Prevention

When it happens

Trigger: Calling `expect(actualResult).toBeSlowerThan(expected)` where `expected` is a plain number, string, raw function, undefined, or any object lacking a numeric `latency.mean`. Both arguments must originate from `bench()` tasks.

Common situations: Passing a millisecond number instead of a bench result object (e.g. `.toBeSlowerThan(100)`), comparing against a pre-saved number, or destructuring a result incorrectly so `latency` is undefined. Also happens when comparing a `bench()` result against a result from a different benchmarking library.

Related errors


AI-assisted analysis of vitest-dev/vitest@d568f8ce37 (2026-08-03). Data as JSON: /data/errors/22e78e909b445919.json. Report an issue: GitHub.