Usage shows the cost by email, app, and model. You can narrow down

• Filter by specific date ranges, e.g. Before 1 Jan 2025
• Filter a specific app, e.g. LLM Foundry Playground
• Filter a specific email, e.g. s.anand@gramener.com

Here are some best practices to manage costs:

1. Start with the cheapest model. See the Model Picker to explore which model to pick and how much it costs. Stay with the cheapest model that works for you.
2. Monitor costs daily at first. Once usage stabilizes, switch to weekly monitoring.
3. Use the LLM Foundry cache. LLM Foundry caches results if the input is exactly the same. Avoid sending slightly different inputs (e.g. standardize templates, temperature, etc.)
4. Review the prompts. Trim unnecessary text. Use shorter instructions. Use system prompts for common instructions.
5. Find better models. Keep an eye on Model Picker for better, newer models that are cheaper and migrate to them.