Remove Released column; rename Tokenizer Tax to Tokenizer Inefficiency

Remove redundant Efficiency column (same signal as Tokenizer Tax, less honest scaling)

Add Tokenizer Tax column; fix pct_more formula to match analyze_benchmark.py

Update script.js

Update script.js

Update script.js

Update script.js

Update script.js

UI: match description to sketch font; show all 14 models in chart

UI: add description, fill chart bars, normalize model-name casing

Add claude-fable-5 & grok-4.5 tokenizer results (script.js)

Upload 4 files