API costs are the second-biggest expense (after compute) for AI startups. Here is a free CLI tool to compare them.
The Problem
You need to pick an AI API provider. Prices vary wildly:
- DeepSeek V4 Flash: $0.15/M tokens
- GPT-5.5: $5.00/M tokens
- Claude Haiku 3.5: $0.80/M tokens
But comparing blends of input/output costs across models at your specific volume is tedious.
The Solution: ai-model-cost
A zero-config CLI tool. Install it in one command:
npx ai-model-cost --tokens=100M --compare
Output shows every major model sorted by price with monthly cost at your volume.
Quick Examples
# Compare all models at 100M tokens
npx ai-model-cost -t=100M -c
# Check a specific model
npx ai-model-cost --model=gpt-5.5 --tokens=500M
# List all available models
npx ai-model-cost --list
The Results at 100M Tokens/Month
| Model | Monthly Cost | vs GPT-5.5 |
|---|---|---|
| DeepSeek V4 Flash (ModelHub) | $18 | 45x cheaper |
| Gemini 2.0 Flash | $22 | 36x cheaper |
| DeepSeek V4 Flash (Official) | $9 | 91x cheaper |
| GPT-5.5 | $800 | baseline |
Bottom Line
Before signing up for a $2,000/month API bill, run this tool. It takes 5 seconds.
Top comments (0)