ACCA - Acidni Chat Cost Analyzer
A VS Code extension that saves you money by analyzing AI token usage - both for instruction files AND your actual GitHub Copilot chat histories.
💰 How It Saves You Money
| Analysis Type |
What It Does |
Savings |
| Chat History |
See what you ALREADY spent |
Identify expensive patterns |
| Quality Analysis |
Compare efficiency across sessions |
Learn what works best |
| Instruction Files |
Optimize prompts BEFORE you use them |
Reduce future costs |
Features
📊 Chat History Cost Analysis
See exactly what your AI conversations are costing you:
- BY COST: Sessions grouped by High/Medium/Low cost
- BY QUALITY: Sessions ranked by efficiency (chars per dollar)
- Per-conversation cost breakdown
- Input vs output token tracking
- Most expensive sessions highlighted
📈 Quality Rating System
Sessions are automatically ranked using percentile-based quality scoring:
| Rating |
Percentile |
Meaning |
| 🌟 High Quality |
Top 33% |
Best value - most output per dollar |
| ⚡ Medium Quality |
Middle 33% |
Average efficiency |
| ⚠️ Low Quality |
Bottom 33% |
Room for improvement |
Quality = Total Characters ÷ Total Cost (higher is better)
🤖 Supported Models
| Model |
Cost Multiplier |
Pricing Included |
| GPT-4o |
1x |
✅ |
| GPT-4o-mini |
0x (free) |
✅ |
| GPT-4.5-preview |
6x |
✅ |
| GPT-4 |
1x |
✅ |
| o1 |
6x |
✅ |
| o1-mini |
1x |
✅ |
| o3-mini |
1x |
✅ |
| Claude 3.5 Sonnet |
3x |
✅ |
| Claude Sonnet 4 |
3x |
✅ |
| Claude Opus 4 |
6x |
✅ |
| Gemini 2.0 Flash |
0x (free) |
✅ |
| Gemini 2.5 Pro |
3x |
✅ |
🔢 Token Counter
- Real-time token count in status bar
- Count tokens in any selection
- Accurate gpt-tokenizer based counting
📋 Instruction Analyzer
Analyzes your Copilot instruction files and categorizes by efficiency:
| Tier |
Format |
Efficiency |
| 1 |
YAML/JSON |
⭐⭐⭐⭐⭐ |
| 2 |
Bullet Points |
⭐⭐⭐⭐ |
| 3 |
Short Sentences |
⭐⭐⭐ |
| 4 |
Prose/Paragraphs |
⭐⭐ |
Installation
- Open VS Code
- Press
Ctrl+P (or Cmd+P on Mac)
- Type
ext install AcidniLLC.ai-chat-cost-analyzer
Or install from VS Code Marketplace
Usage
View Chat Costs
- Open the ACCA panel in the sidebar
- Expand BY COST to see sessions grouped by expense
- Expand BY QUALITY to see efficiency rankings
- Click any session for detailed analysis
Analyze Instructions
- Open your
.github/copilot-instructions.md
- Run command:
ACCA: Analyze Instructions
- View efficiency report with suggestions
Count Tokens
- Select text in any file
- Run command:
ACCA: Count Tokens
- See token count in notification
Commands
| Command |
Description |
ACCA: Scan Chat Sessions |
Refresh chat session analysis |
ACCA: Analyze Chat History Costs |
Detailed cost breakdown |
ACCA: Show Chat Cost Report |
Full spending report |
ACCA: Open Cost Dashboard |
Visual analytics dashboard |
ACCA: Scan Instruction Files |
Find and analyze instruction files |
ACCA: Count Tokens |
Count tokens in selection |
Settings
{
"chatCostAnalyzer.tokenModel": "gpt-4o",
"chatCostAnalyzer.showStatusBar": true,
"chatCostAnalyzer.pricing": {
"gpt-4o-input": 0.0025,
"gpt-4o-output": 0.01
}
}
Why Optimize Instructions?
Prose (inefficient):
When writing code for this project, please make sure to use TypeScript
with strict mode enabled. It would be great if you could prefer using
const declarations over let whenever possible.
~45 tokens
Bullet Points (efficient):
- Use TypeScript strict mode
- Prefer const over let
~12 tokens
Savings: 73% fewer tokens = lower cost + better AI compliance
License
MIT © Acidni LLC
Contributing
Issues and PRs welcome at GitHub
Made with 💚 by Acidni LLC
