Evaluates whether a model can count occurrences of the lowercase letter “r” in words and word-like strings while returning only the correct integer.
Scorer: ExactMatch
loading chart
| # | score | model | cost price | ttft | total tok |
|---|---|---|---|---|---|
| 1 | 90% | anthropic/claude-haiku-4.5 | $0.00007 $1.00 / $5.00 per 1M | 1.56s | 50.7 |
| 2 | 70% | gpt-5.2 | $0.00013 $1.75 / $14.00 per 1M | 0.65s | 45.3 |
| 3 | 70% | anthropic/claude-opus-4.5 | $0.00035 $5.00 / $25.00 per 1M | 1.44s | 50.7 |