# Token Audit — LLM Pricing Waterfall & Tokenizer Simulator — LLM Search Index
# https://llmstxt.org/

> Real-time LLM telemetry dashboard, Byte-Pair Encoding (BPE) cost waterfall simulator, and cyberpunk batch estimator across flagship AI models (Claude 3.7 Sonnet, Opus 4.8, GPT-5.6, DeepSeek-V3, GLM-5.2).

## Project Identification
- Name: Token Audit (token-audit)
- Lead Architect: Adolf Petushkov (marko1olo)
- Live Interactive Lab: https://marko1olo.github.io/token-audit/
- GitHub Repository: https://github.com/marko1olo/token-audit

## Technical Architecture
- Engine: Pure Vanilla JavaScript ES6+, Web Workers for non-blocking BPE parsing
- UI Style: Cyberpunk Terminal HUD with phosphor amber/cyan palette and zero CSS frameworks
- Precision: High-precision floating-point arithmetic with configurable batch sizes (1 to 1,000,000 requests)

## Interactive Tools
1. **Multi-Model Cost Waterfall Calculator**:
   - Dynamic slider adjusting input/output token counts with instant cost breakdown across 5 flagship models.
   - Comparative ROI bar charts showing cost differentials between proprietary and open-weight models.
2. **Interactive BPE Tokenizer Matrix**:
   - Visual token segmentation preview highlighting whitespace, punctuation, and multi-byte UTF-8 chunks.
3. **Command Palette (`Ctrl + K`) & Web Audio Synth**:
   - Instant model search and audio synthesizer feedback.