Random Generator Guide: Numbers, Strings, Colors, Dice & UUIDs Online (2026)

Character Sets

Entropy by Length

For passwords: 12+ characters from full ASCII set (78+ bits entropy). For API tokens: 32+ alphanumeric characters (190+ bits).

UUID v4 = 122 random bits = 5.3 × 10³⁶ possible values.

Collision Probability

Generating 1 billion UUIDs per second for 100 years: probability of a single collision ≈ 0.00000000006%. For all practical purposes: zero.

When to Use UUIDs

• Database primary keys — no central counter needed, safe for distributed systems
• Session tokens — unguessable identifiers
• File names — unique upload filenames, no collision with existing files
• Idempotency keys — prevent duplicate API operations
• Event tracking — unique IDs for analytics events

UUID Format

550e8400-e29b-41d4-a716-446655440000
^^^^^^^^-^^^^-4^^^-[89ab]^^^-^^^^^^^^^^^^
         (v4 marker) (variant)

Test Data Generation

• Load testing — Random user profiles, addresses, transactions for API stress tests
• UI testing — Random string lengths to test layout overflow, edge cases
• Database seeding — Realistic but randomized data for development environments
• Fuzz testing — Random inputs to find crashes and security vulnerabilities

Code Examples

// JavaScript — random integer in range
const randInt = (min, max) => Math.floor(Math.random() * (max - min + 1)) + min;

// JavaScript — cryptographic random string
const randString = (len) => {
  const arr = crypto.getRandomValues(new Uint8Array(len));
  return Array.from(arr, b => b.toString(36).padStart(2, '0')).join('').slice(0, len);
};

// Python — random choice from list
import random
random.choice(['red', 'blue', 'green'])

// Python — cryptographic token
import secrets
secrets.token_urlsafe(32)  # 32-byte URL-safe token

Modulo Bias

Mapping a large range to a smaller one using modulo (%) creates slight bias. Example: rand() % 6 where rand() returns 0-7 gives 0 and 1 a higher probability (2/8) than 2-5 (1/8). Modern libraries handle this — but raw modulo on crypto random bytes has this issue.

Fisher-Yates Shuffle

The only correct way to shuffle an array. array.sort(() => Math.random() - 0.5) is biased — some permutations are more likely than others. Use Fisher-Yates:

for (let i = arr.length - 1; i > 0; i--) {
  const j = Math.floor(Math.random() * (i + 1));
  [arr[i], arr[j]] = [arr[j], arr[i]];
}

Common Mistakes

• Using Math.random() for security — predictable, not cryptographic
• Same seed in production — reproducible sequences are exploitable
• Expecting no streaks — 5 heads in a row is normal randomness, not a bug
• Small sample conclusions — 10 rolls averaging 4.2 instead of 3.5 does not mean the generator is biased