Source Evidence Insight (@jun_song): "The global median annual income is only around $5,000... at least 4 billion people make less than $450 a month." Non-Latin writing systems suffer up to 3-5x subword fragmentation overhead in standard LLM tokenizers, amplifying AI access cost inequality.
Global Baseline Affordability
0.034%
of $5,000 Global Median Income
Max Token Bloat Ratio
3.82x
Hindi (Devanagari) vs English
Selected Monthly Usage
500k
Tokens / Month Baseline
Avg Output Cost / 1M
$0.28
Custom Model Tier
⚙️ Token & Pricing Parameters
📊 Tokenizer Script Fragmentation Matrix
Comparing subword overhead against English baseline| Language | Script | Char Count | Est. Tokens | Bytes / Token | Cost / 100k Queries | % Median Income |
|---|