> ## Documentation Index
> Fetch the complete documentation index at: https://docs.opsbeacon.co/llms.txt
> Use this file to discover all available pages before exploring further.

# AI Efficiency Scores

> A scored view of how efficiently each project converts token spend into useful output.

## What the efficiency score measures

Each AI project gets a score from 0–100 reflecting how efficiently it uses tokens. The score factors in:

* Token waste from repeated or failed requests
* Prompt-to-completion ratio vs benchmarks
* Cost per unit of output vs similar workloads
* Cache hit rate (for providers that support prompt caching)

## How to use it

```
@Beacon what's the AI efficiency score for the data-extraction project?
@Beacon show me efficiency scores for all AI projects
@Beacon which AI project has the most room to improve?
```

## Improving a low score

A low efficiency score usually points to one of three root causes:

1. **Prompt bloat** — prompts that are longer than they need to be
2. **High retry rates** — requests failing and being retried at full cost
3. **No caching** — identical or near-identical prompts being sent repeatedly without prompt caching


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.