Are AI Leaderboards Still Worth Taking Seriously? Rankings Disagree Daily
Snappy-Domino4017 · reddit · 2026-11-15
A Reddit user questions the value of AI leaderboards: every new benchmark crowns a different "best in the world" model, and rankings flip depending on the site — Gemini leads general chat, Claude leads coding.
High benchmark scores don't prevent garbage answers to basic questions, and Kalshi even runs a prediction market on which model tops November. The author trusts practitioners' daily experience over vendors announcing 3-point wins, and asks whether anyone has switched their main AI because of rankings.
More from Models
- Hyped as AGI, Argon Feels Worse Than Gemini 3.8 Flash in Early Use — ZealousidealNeat2246 · 2026-11-15
- NVIDIA open-sources SkillEvaluator, a multi-tier framework for evaluating AI agent skills — NVIDIA · 2026-11-15
- Anthropic says start most coding work on Opus 5.5, not Fable 5.2 — effort beats model size — blaizedsouza · 2026-11-15
- Same savings app prompt: Claude vs ChatGPT vs Figma Make UI showdown — Tegadesigns · 2026-11-15
- Epoch ECI Trend Fit: Anthropic's Latest Model Runs ~5 Months Ahead — peterwildeford · 2026-11-15
- Report: Gemini silently drops the first two-thirds of saved custom instructions — East_Dimension4778 · 2026-11-15