Your LLM budget alerts won't save you if you can't map costs to users

Source: Dev.to

The Problem with Hard Caps

Most developers think they have AI costs under control because they set a $500 hard cap on OpenAI.
In reality, that’s not cost control—it’s just a timer for when your app goes down.

When you hit $400 and receive a warning email, you look at the dashboard and see a massive spike in GPT‑4o usage. But who caused it?
The provider dashboard won’t tell you; it only shows a giant wall of tokens.

If you’re building a multi‑tenant SaaS, you need to map every single LLM call to a specific user ID. Without attributing the cost, you can’t understand your unit economics and might be losing money on your power users.

Mapping Costs to Users

LLMeter is an open‑source dashboard that sits between your app and the provider. By passing the user ID in the metadata, LLMeter tracks the cost per user, per day, across:

• OpenAI
• Anthropic
• DeepSeek
• OpenRouter

Features

• No proxies, no Stripe lock‑in – just raw data.
• Self‑hostable under AGPL‑3.0.

Check it out here: