How are
vibes today?
VibeBench is a live, community-powered leaderboard that tracks the real-time vibe of popular LLM models.
Rank
Model
Vibe Score
Curse Trend
Vote
Live feed
Last updated: 2:34 PM
TIME
MODEL
RATING
How it works
The basics
VibeBench runs on community votes (Fire, Mid, Cursed) to calculate real-time Vibe Score and Curse Trend for popular LLM models.
Both metrics use statistical normalization so models with only a few votes can't game the leaderboard.
The basics:
- •Vibe Score - 0-100 rating for selected time period (higher is better)
- •Curse Trend - Shows % cursed votes per hour. Helps spot degraded performance
- •Voting: - You get 3 votes per hour, max 1 per model
Community driven
More votes = better data. So next time a model deletes half your codebase, after you're done raging in all caps, come curse it on VibeBench (or praise it when deserved).
Prefer the terminal?
zsh
$vibebench vote gpt-4o cursed 'deleted my entire function'
✓ Vote recorded for gpt-4o
$vibebench stats claude-4.1-opus
claude-4.1-opus
Rank: #4
Vibe Score: 63.4/100
Votes: 🔥 47 😐 23 💀 12
$_