AiGanakFind out where your LLM bill goes, and what you can save.
Drop in your OpenAI or Anthropic usage export. In seconds you see spend by model, your monthly run rate and the savings from model routing, prompt caching and batch processing. Your file is never uploaded.
A free, private tool that reads an OpenAI or Anthropic usage export privately on the user's own computer, shows spend by model and the monthly run rate, and estimates savings from model routing, prompt caching and batch processing. The file is never uploaded.
No file to hand?
Paste a table instead
Columns we detected · change if anything looks wrong
Spend by model
| Model | Cost | Share | Tokens |
|---|
Your savings estimate
Savings are applied in order (routing, then caching, then batch) so nothing is counted twice. Routing assumes the same provider's mid-size model at list prices checked 24 October 2026; it only holds if quality stays the same on your own examples.
Turn the estimate into a plan
The full LLM Cost Report analyses a sample of your real requests, measures what caching can cover, prices your workload on every current model and gives you a ranked plan with code for your provider. USD 149 launch price (normally 199), delivered in two working days.
Need cheaper models tested on your own answers? The LLM Cost & Reliability Audit goes further.
Choose how deep you want to go
Free tools
- Spend by model and monthly run rate
- Every major model priced for your traffic
- Savings estimate from your assumptions
- Private: your file is never uploaded
Sample report
- Your numbers in a branded report you can share
- Your workload priced on every current model
- Savings split by routing, caching and batch
- What the full report would add for you
Full LLM Cost Report
- Request-level analysis of your real calls
- Cacheable prompt prefixes measured, not guessed
- Ranked savings plan with code for your provider
- Which changes need a quality test, and how
- Three-month cost forecast, including announced price rises
LLM Cost & Reliability Audit
- Everything in the full report
- Cheaper models tested on your own approved answers
- Prompt and reliability review
- Walkthrough and implementation plan
How to export your usage
OpenAI
- Open platform.openai.com, pick the organisation and project.
- Go to Usage and set the date range, for example last month.
- Click Export, choose Cost data, group by Line item, format CSV, interval 1 day.
- Download the file and drop it above.
Activity data exports with token counts also work.
Anthropic
- Open the Claude Console and go to the usage or cost page.
- Set the date range and export as CSV.
- Drop the file above. If a column is not recognised, pick it in the column list.
Exports from Azure, Bedrock, Gemini or your own logs work too, as long as they have a cost column or token columns.
Your file never leaves your computer
Your file is analysed on your own computer and is never uploaded. There is no account and no tracking. If you ask for a sample report by email, only the headline totals are sent.
Want current list prices for every model? See the LLM API pricing calculator.
Bill Analyzer questions
Can't find it? Press ⌘K and ask.
Is my usage file uploaded anywhere?
No. The analysis happens on your own computer. The file is never sent to AiGanak or anyone else. Only if you ask for a sample report by email are the headline totals sent.
Which files does the Bill Analyzer accept?
Any CSV usage or cost export: OpenAI cost data grouped by line item, OpenAI activity data, Anthropic Console exports, and exports from Azure, Bedrock, Gemini or your own logs, as long as there is a cost column or token columns. You can tell it which column is which.
How are the savings estimated?
Three changes are applied in order so savings are not counted twice: moving part of flagship-model traffic to the same provider's mid-size model, covering part of uncached input with prompt caching (about 90% cheaper), and moving work that can wait to half-price batch processing. You set each share.
How accurate is the estimate?
It is a first estimate from your billing data and your assumptions. Real savings depend on whether a cheaper model keeps quality on your tasks and how much of your prompts can be cached, which is what the full LLM Cost Audit tests on your own approved answers.