Character Frequency Bench

Tests an LLM's ability to accurately count and categorize specific characters, symbols, and patterns within strings. This benchmark evaluates tokenization-independent visual processing and precise sub-string analysis.

Jan 9, 2026
11 tasks
110 models
$0.7334
karllorey
Link only

Tests

Each test is one prompt sent to every model in the benchmark.

11 tests × 110 models = 2420 arena votes for reliable rankings.