Projects

model evaluation

chatio benchmark

A leaderboard that evaluates LLMs on real tasks instead of abstract puzzle scores.

Evaluation design, visual system, full-stack engineering2025

What I built

The benchmark scores models across advice, instruction following, reading comprehension, empathy, and creative writing. The public leaderboard also shows price, context length, output limits, provider, and response speed so the score has practical context.

Selected capabilities

  • Five evaluation areas focused on practical usefulness
  • Searchable rankings with provider and model filters
  • Methodology published alongside model scores
Claude Opus 4.5
claude-opus-4-5-20251101
0.0
Helpfulness
Instruction Following
Comprehension
Empathy
Creative Writing
Cost per 1M$5.00 in
$25.00 out
Context Size200,000 Tokens
Max Output Tokens64,000 Tokens
Model Speed60 Tok/Sec
Interactive model picker
Chatio Benchmark model ranking and score details
Rankings, evaluation scores, pricing, and model limits