DEVELOPER SENTIMENT / NOT A BENCHMARK
AI confidence.
Priced by developers.
How derpy is your AI today? A reality check from the people in the terminal.
CONNECTING
COMMUNITY DERP SCORE PAST 24H
Per agent · pick one for its modelsHigher = more facepalms. Empty samples have no score.
Sentiment, not sales pitches.
Derp Score over time. Higher means derpier. Every vote has a say.
Claude CodeOpenAI CodexGemini CLI
AGENT / PROVIDERDERP SCORE / 7D24H CHANGE7D SAMPLEYOUR VERDICT
Anthropic
— No baseline
—Loading
OpenAI
— No baseline
—Loading
Google
— No baseline
—Loading
Anonymous. Server-side rate limits apply. DERP SPIKE WATCHToo fewToo fewToo few≥30 current / ≥100 baseline votes Rating mix
No votes in this window.
Why it went wrong
No Meh or Dumb votes in this window.
Derp Spike detail
Latest 24h vs the prior 7 days. A spike needs a 10+ point rise that is also statistically significant, from at least 10 networks.
By model
No model-level votes yet.
By channel
Session-linked vs public
Session-linked means an opted-in local agent session was observed. It is not verification.
Release impact
3 days before vs after the release day (day excluded). Correlation, not causation.
Release radar
VERIFIED DATESNew model. New promises. See what changed.
A curated selection of source-verified releases. Dates are not exact launch times. Correlation does not establish causation.
DERPMETER EVERYWHERE
Submission routes
Where votes were submitted, separate from the AI agent being rated.
Web
Dashboard voting
MCP
Claude Code · Codex
Menu Bar
macOS companion
CLI
Direct terminal votes
Share of each agent's own votes by channel. Independent of chart filters; client labels do not verify model usage.
