CostFunctionBack to Home

User Guide

Everything you need to get started with CostFunction.

1. Uploading Images and Running an Analysis

CostFunction lets you compare the cost and quality of AI image analysis across multiple providers and models.

  1. Log in to your account and navigate to the Dashboard.
  2. Click Upload Images to select one or more images from your device. Supported formats include JPEG, PNG, and WebP.
  3. Choose the AI models you want to compare. You can select multiple models to run side-by-side.
  4. Click Run Analysis. CostFunction will send your images to each selected model and collect the results.

Analysis typically completes within a few seconds per model. You can track progress in real time on the results page.

2. Understanding Your Results

After an analysis completes, you will see a detailed report containing:

  • Model Comparison — Side-by-side quality scores and response details for each model you selected.
  • Cost Projection — Estimated costs per image and at scale, helping you forecast expenses for production workloads.
  • Recommendation — A suggested best-value model based on quality-to-cost ratio.
  • Sample Analysis — The actual AI-generated descriptions for each image so you can evaluate output quality directly.

You can download a PDF report for offline review or sharing with your team.

3. Credits and Billing

CostFunction uses a credit-based system for Pay-As-You-Go users:

  • 1 credit = 1 analysis report. Each credit lets you run a full multi-model comparison on your images.
  • Credits never expire. Use them whenever you need.
  • You can purchase additional credits from your Billing page at any time.
  • Credits are only deducted after a successful analysis (at least 80% of models must return results).

Subscription plans (Full Service and Developer BYOK) include unlimited analyses — no credits required.

4. Developer (BYOK) Setup

If you are on the Developer (BYOK) plan, you need to add your own API keys before running analyses:

  1. Go to Settings → API Keys.
  2. Add API keys for the providers you want to use (e.g., OpenAI, Anthropic, Google).
  3. Your keys are encrypted with AES-256 and stored securely. We recommend using dedicated keys with spending caps — never use your production keys.
  4. Once configured, those providers will appear as available models when you start an analysis.

5. Getting Help

If you run into any issues or have questions: